ΠΑΝΕΠΙΣΤΗΜΙΟ ΠΑΤΡΩΝ ΣΧΟΛΗ ΕΠΙΣΤΗΜΩΝ ΥΓΕΙΑΣ ΤΜΗΜΑ ΙΑΤΡΙΚΗΣ

Transcription

1 ΠΑΝΕΠΙΣΤΗΜΙΟ ΠΑΤΡΩΝ ΣΧΟΛΗ ΕΠΙΣΤΗΜΩΝ ΥΓΕΙΑΣ ΤΜΗΜΑ ΙΑΤΡΙΚΗΣ «Υπολογισµός πιθανότητας µετακτινικής επιπλοκής µε βελτιστοποίηση ακτινοβιολογικών παραµέτρων σε ασθενείς µε καρκίνο του πνεύµονος και συσχέτιση µε δοκιµασίες αναπνευστικής λειτουργίας.» ΣΒΩΛΟΥ ΠΑΤΡΙΤΣΙΑ ιπλωµατική Εργασία Αρ.Μητρώου: 1288 Εξεταστική Επιτροπή: Καθηγητής Κ. Κάππας Επ. Καθηγήτρια Κ. Θεοδώρου Επ. Καθηγήτρια Ε. Κωσταρίδου Επιβλέπων: Καθηγητής Κ. Κάππας 1

2 UNIVERSITY OF PATRAS SCHOOL OF MEDICINE MEDICAL PHYSICS DEPARTMENT «Estimation of radiation induced complications probabilities with radiobiological parameter optimization in lung cancer patients and correlation with pneumonological function tests» SVOLOU PATRICIA Msc Thesis I.D Νο : 1288 Examination Committee: Prof. C.Kappas Assoc. Prof. K. Theodor Assoc. Prof. Ε. Costaridou Supervisor: Prof. C.Kappas 2

3 3

4 Acknowledgments I would like to thank my Professor Constantine Kappas for his support and help, for our priceless conversations and for passing on his passion on Medical Physics. I would also like to thank my supervisor Assistant Professor Kiki Theodorou for her scientific guidance, her help and support throughout the course of this present study and for her friendship. I would like to express my unlimited gratitude to Dr. Ioannis Tsougkos for his excellent supervision, for guiding and supporting me when things got rough, for his help and most important for his friendship. Furthermore, I would like to thank my colleagues in the Medical Physics Department, at the University Hospital of Larisa. Last but not least, a special thanks to my family and friends who are supporting me throughout these years, for their constant encouragement and for being there in difficult times. 4

5 Abstract The treatment techniques applied in the chest area (breast and lung cancer) in radiotherapy, increase the lung toxicity leading to the development of pulmonary complications. The aim of this study is to compare the predictive strength of different radiobiological models in the evaluation of radiation pneumonitis, correlate the absorbed dose with the severity of the clinical outcome and examine biological factors that may affect or induce complications after irradiation. Furthermore, due to the fact that the value of each parameter is accompanied by its confidence interval, every model is represented by a group of dose-response curves that create a range in which each radiobiological model can vary. The range of each model is very important when selecting the values of parameters used, due to the existence of coincendence areas between the models. The study was based on 179 breast cancer patients undergoing radiotherapy. Dose volume histograms and the clnical treatment outcome for every patient were available. 24 patients scored radiation pneumonitis grade 2 and 65 showed milder symptoms of grade 1. Normal tissue complication probabilities were computed for every model for each patient. Moreover, statistical analysis was applied to investigate whether the absorbed dose is the only factor that influeces the development of radiation pneumonitis in breast cancer radiotherapy (x 2 test) and the ability of the radiobiological models used to discriminate cases that developed radiation pneumonitis from those that did not (ROC curves). The Relative Seriality model described with greater accuracy the clinical outcome in contrast to the LKB and Parallel model. The statistical results showed that radiation pneumonitis in the case of breast cancer radiotherapy does not depend only on the absorbed dose but on other radiobiological factors that induce the development of complication, such as the intrinsic radiosensitivity of each patient. ROC curves pointed out the weakness of the models to discriminate cases of complication from cases of non-complication. Finally, this study accented the importance to use parameter values extracted from patients groups with similar clinical characteristics as the ones examined in order to avoid the coincidence areas between the models. 5

6 Contents Acknowledgments... 4 Abstract Introduction The role of radiotherapy in cancer management The role of Radiobiology Biologic Effect of Radiation Linear Energy Transfer (LET) Relative Biological Effectiveness (RBE) Radiation effect on cells Biological Impact of Radiation Cell Cycle DNA Structure DNA Damage Radiation effects on tissues Early-Late Responding Tissues The role of time factor Cell survival models Cell Survival Curves- LQ Model Biologic Effect of Parameters a, b Biological significance of a, β ratio Radiosensitivity Fractionation Radiobiological approach and models Radiobiological approach in Radiotherapy optimization Importance of radiobiological approach in radiotherapy Radiobiological models Biological significance of the radiobiological parameters in the models Dose-Response curves Poison model NTCP Models Methods and Materials PART I Purpose of project Patient group- Method of treatment Dose Volume Histogramms

7 2.4 Radiobiological Models Scoring Normal Tissue Complications Statistical Analysis PART II Protocol: Estimation of Radiation Pneumonitis based on physical and biological parameters in lung cancer patients undergoing radiotherapy RESULTS DISCUSSION CONCLUSION REFERENCES

8 1. Introduction 1.1 The role of radiotherapy in cancer management. Radiation therapy is one of the three principal modalities used in the treatment of malignant disease (cancer), the other being surgery and chemotherapy. In contrast to other medical specialties that rely mainly on the clinical knowledge and experience of medical specialists, radiotherapy, with the use of ionising radiation in cancer treatment, relies heavily on modern technology and collaborative efforts of several professionals whose coordinated team approach influences greatly the outcome of the treatment. Radiotherapy has replaced surgery for the long-term control of many tumors of the head and neck, cervix, bladder, prostate and skin, in which it often achieves reasonable probability of tumor control with good cosmetic results. In addition to these examples of the curative role of radiation therapy, many patients gain valuable palliation by radiation. The use of the modern linear accelerator has become a very precise tool, capable of depositing a defined dose to a specific volume of tissue, rendering radiotherapy an increasingly important modality for the treatment of most cancers. This has been made possible by rapid advances in technology, including intensity modulation and image guidance in real time. These developments have been particularly useful in allowing sparing of normal tissues lying in close proximity to tumors, such as bowel adjacent to cancer of the prostate. Chemotherapy is the third most important treatment modality at the present time. Following the early use of nitrogen mustard during the 1920s, it has emerged to the point where upwards of 30 drugs are available for the management of cancer, although no more than are in common use. Many patients receive chemotherapy at some point in their treatment and useful symptom relief and disease arrest are often obtained. Very substantial numbers of patients with common cancers achieve long-term tumor control largely by the use of radiation therapy. Local treatment which includes surgery and/or radiotherapy, could be expected to be successful in approximately 40% of these cases, in perhaps 15% of all cancers, radiotherapy would be the principal form of treatment. By contrast, many patients do receive chemotherapy but their contribution to the overall cure rate of cancer may be only around 2%, with some prolongation of life in perhaps another 10%. This is because the diseases in 8

9 which chemotherapy does well are rare. The important benefits of chemotherapy in a number of chemosensitive diseases should not be undervalued, but to stress the greater role of radiotherapy as a curative agent. Considerable efforts are being devoted at the present time to the improvement of radiotherapy. There are three main ways in which such an improvement in radiotherapy might be obtained: By raising the standards of radiation dose prescription and delivery to those currently in use in the best radiotherapy centers. By improving radiation dose distribution beyond those that are conventionally achieved, either using techniques of conformal radiotherapy with photons or ultimately by the use of proton beams. By exploiting radiobiological initiatives. The above developments will have an excess impact on treatment success in the future. [2] The role of Radiobiology Experimental and theoretical studies in radiation biology contribute to the development of radiotherapy at many levels. First of all, radiation biology offers a fundamental basis in the understanding of how radiotherapy works. It identifies the mechanisms and processes that influence tumor and normal tissue response to irradiation, and make us able to understand and explain the observed phenomena. Obvious examples of this knowledge are the understanding of hypoxia, reoxygenation, tumor cell repopulation or mechanisms of repair of DNA damage. Moreover, radiation biology has led scientists to approach radiotherapy in a different way, by recommending cell sensitizers, high-let radiotherapy, accelerated radiotherapy and hyperfractionation. According to the above, it is clear that through the knowledge of the processes that conceptualize the tissue response to radiation, scientists have pointed to the advice on the choice of schedules for clinical radiotherapy. For example, conversion formulae for changes in fractionation or dose rate, or advise on whether to use chemotherapy only or a combination of chemotherapy and radiation. Furthermore, great effort has been made in another arising field of radiobiology, that 9

10 of predicting the best treatment for individual patient (individualized radiotherapy). The newer conversion formulae based on the linear-quadratic equation seem to be very successful. Beyond this though, these theoretical and experimental formulae are still inadequate in the decision of the specific protocols that govern a clinical practice. Taking into consideration all the above, there is no doubt that radiobiology has been very fruitful in the generation of new ideas and in the comprehension of potentially exploitable mechanisms. Unfortunately, through many different treatment strategies, very few are being demonstrated in clinical practice. [2] 1.2 Biologic Effect of Radiation Linear Energy Transfer (LET) Linear Energy Transfer is defined as followed: LET of a charged particle in a medium is the quotient de/dl, where de is the average energy locally imparted to the medium by a charged particle of specified energy in traversing a distance of dl. In contrast to the stopping power with a typical unit of MeV/cm, the unit usually used for LET is kev/µm. The energy average is obtained by dividing the particle track into equal energy increments and averaging the length of track over which these energy increments are deposited. Figure 1 Linear Energy Transfer The LET factor represents the energy loss rate by collisions of the secondary electrons per unit path length so the energy is absorbed locally. The ability of radiation to transfer its energy to matter, inducing biological reactions is determined 10

11 through LET. So LET, is of particular importance when the particles pass through living tissue as it modifies the effect of a specific dose of radiation. It is proportional to the square of the particle's charge and increases as the velocity of the particle decreases. Figure 2 LET as a function of particle speed Charged particles, such as alpha, cause far more biological damage per quanta than photons of the same energy do, because they deposit their energy on shorter distances inside the tissue. Alpha particles may deposit energy at rates up to 1000 times higher than electrons. High-LET particles produce stronger biological effects than low-let particles do. The varying biological effectiveness, or the potential one, for different types of radiation is introduced through quality factor Q. Q varies with LET, and is quantified for a particular radiation by comparing the dose of the radiation required to produce a particular biological effect with the dose of a reference radiation required to elicit the same effect. The reference radiation may be medium energy (e.g. 200-keV) X rays or 60Co radiation. Dose equivalent, H, expresses the damaging effects of different radiations on a common scale by incorporating the quality factor: (1.96) The Sievert is the SI unit for dose equivalent. An older unit still in common use is the rem. Quality factors for several types of radiation are listed in Table 1 In the field of radiation biology, a similar modifying factor, the relative biological effectiveness or RBE, is used to compare the potential for damage of various types of radiation on a common scale. [3] 11

12 Table 1 Quality factors for different types of radiation From Table 1 we can see that neutrons are described as high-let radiation even though they are uncharged. They do not interact with the orbital electrons as they pass through tissue and they do not directly produce ionization. They do, however, interact with atomic nuclei from which they eject slow, densely ionizing protons. It is this secondary production of knock-on protons that confers high- LET. [2] LET of different types of radiation Type of radiation LET (kev/µm) 25 MeV X-rays MeV electrons 0.3 Diagnostic X-rays MeV protons 4.0 Fast neutrons MeV alpha particles Heavy nuclei Table 2 LET for different types of radiation High value of L.E.T (in KeV/µm) indicates the presence of ionizations taking place in high density. So radiations with high L.E.T. value cause the double DNA strand to break. The breaking of the double DNA strand is very difficult to be repaired causing the total destruction of the cell. Radiations having L.E.T.<10 KeV/µm are considered as low L.E.T.s, whereas radiations with L.E.T.>10 KeV/µm are considered as high L.E.T.s [4] X-rays and gamma rays are considered low LET (sparsely ionizing) radiations, while energetic neutrons, photons and heavy charged particles are high 12

13 LET (densely ionizing) radiations. The demarcation value between low and high LET is about 10 kev/µm Relative Biological Effectiveness (RBE) The effect of radiation on living organisms depends strongly on the LET of radiation as well as on the biological system itself. As the LET of radiation increases, the potential biologic damage also increases. Taking into account that different types of radiation have different LETs, e.g neutrons and X-rays, they induce different biological damage. This difference is based on the pattern of energy disposition. Radiations of different qualities have different degrees of effectiveness in producing effects in biological systems. When radiation is absorbed in biological material, the energy is deposited along the tracks of charged particles in a pattern that is characteristic of the type of radiation involved. After exposure to x or gamma rays, the ionization density would be quite low. After exposure to neutrons, protons, or alpha particles, the ionization along the tracks would occur much more frequently, producing a much denser pattern of ionizations. Figure 3 RBE for different types of radiation These differences in density of ionizations are a major reason that neutrons, protons, and alpha particles produce more biological effects per unit of absorbed radiation dose than do more sparsely ionizing radiations such as x rays, gamma rays, or electrons. Other factors that contribute to these differences include the energy of radiation used, the dose received and the temporal pattern in which it was received, and the particular biological endpoint being studied. Many scientific investigations 13

14 have been conducted to study the differing effectiveness of radiations under different experimental conditions. In order to compare and contrast the biological effects on tissues, Relative biological effectiveness (RBE) is introduced. RBE compares the dose of test radiation to the dose of standard radiation to produce the same biological effect. The standard radiation is usually taken as 250 kvp x rays for historical reasons. RBE is defined by the following ratio [4] : In general, RBE varies not only with type of radiation but also with type of cell or tissue, biologic effect under investigation, dose rate and fractionation. As the L.E.T. increases the slope becomes steeper and the extrapolation number, n, tends towards unity. RBE versus LET (Fig.4) shows that the relationship increases slowly at first, then more rapidly at LET values beyond about 10keV/µm. The response reaches a maximum at LET values of about 200 kev/µm, and then decreases. The maximum values of 200 kev/µm is similar for a wide variety of mammalian cells and for different endpoints (mutation, cell killing). It is likely that this reflects the target size and is related to the DNA content, which is similar for all mammalian cells. Figure 4 RBE against LET. The vertical dashed line separates the low LET region where RBE 1 from the high LET region where RBE first rises with LET, reaches a peak of about 8 for LET 200 kev/µm and then drops with a further increase in LET 14

15 An increase in the RBE offers no therapeutic advantage, unless there is a differential effect making the RBE for normal tissue smaller than that for the tumour, increasing the relative level of tumour cell killing and the therapeutic ratio. [4] 1.3 Radiation effect on cells Biological Impact of Radiation Biologic damage resulting from ionizing radiation occurs on three levels: molecular, cellular, and organic. The origins of cellular and organic biologic damage begin at the molecular level. Therefore, to understand the biologic effects of radiation on humans, we must first look at its effects on atoms and molecules. Approximately, 80% to 85% of the body is made up of water, so it is likely to be a target of radiation interaction. Figure 5 Body chemical composition Ionization is the process of removing or adding an extra electron to an atom, thereby giving it a negative or positive charge. When an X-ray photon interacts with and ionizes a molecule of water, it dislodges an electron from its orbit. During this process an ion pair is created. When the electron is knocked out of the atom, the molecules charge changes from neutral to positive. The dislodged electron can freely 15

16 interact with another atom. Both parts of this ion pair, the positively charged water molecule (HOH + ) and the negatively charged electron (e-), are unstable. Figure 6 Water molecule ionization If the positively charged water molecule HOH + recombines with the ejected electron, no biologic damage occurs. If the ejected electron joins another H 2 O water molecule, an unstable, negatively charged water molecule is created (HOH - ). This instability can cause the HOH - molecule to dissociate into smaller compounds. The HOH - molecule splits into two components: a hydrogen radical (H*) and a negatively charged hydroxyl ion (OH - ). Free radicals are molecules that have no charge, with equal numbers of protons and electrons, but have a single electron in the outer shell. This means that free radicals are highly chemical active. The positive charged water molecule due to its instability breaks down to produce a hydrogen ion (H + ) and a hydroxyl (OH) free radical. Figure 7 Production of ions and free radicals in water radiolysis Hydrogen and hydroxyl ions can cause minimal biological damage, in comparison with free radicals, because they can easily recombine to form a molecule of water. The unpaired electron in the outer shell allows free radicals to interact with a variety 16

17 of other molecules and destroy chemical bonds by transferring their excess energy to the molecules, creating point lesions. A point lesion is a cellular impairment or loss of a function. Figure 8 Biological impact of water radiolysis Free radicals may react with essential macromolecules. The most important reactions are those with DNA molecules. Damage to DNA from free radicals produced in water is called indirect effect of radiation; ionization of atoms that are part of the DNA molecule is the direct effect (Fig. 9). Since range of free radicals is of few mm, there has been some debate as to whether effects in water that is tightly bound to DNA are direct or indirect. Despite this, it is assumed that 85% of the biological effects are due to indirect hits and only 15% to direct ones. If the damage is not going to be repaired, the reproduction or the survival of the cell is suppressed and it is not unlikely to come up with a living but genetically modified cell. Some organs resist to great loss of cells, but if the number of cells to be lost is much bigger Figure 9 Direct and Indirect effects of radiation than an organ can handle with, then this organ suffers a malfunction. 17

18 In low doses, the probability of a malfunction to be occurred is very low. On the other hand as the dose increases this probability steps up too, and in very high doses tends to be 100% Cell Cycle Proliferating cells undergo a cycle of division, the cell cycle, which lasts approximately 24 hours in mammalian cells in cell culture. The cell proliferation cycle is defined by two well-defined time periods: i) Mitosis M where division takes place, and ii) the period of DNA Synthesis S (Fig.10). The S and M portions of the cell cycle are separated by two periods (gaps) G1 and G2 when DNA is not yet synthesized but other metabolic processes take place. Fully differentiated animal cells only divide rarely. These cells are in the so-called G 0 phase, in which they can remain permanently. Some G 0 cells return to the G1 phase again under the influence of mitogenic signals (growth factors, cytokines, tumour viruses, etc.), and after crossing a control point (G1 to S), enter a new cycle. G1 occurs just after the cell has split. During this phase, the cell begins to manufacture more proteins in preparation for division. It also experiences other growth: metabolism increases, RNA synthesis is elevated and organelles duplicate. Figure 10 Cell proliferation cycle 18

19 During the S phase, the DNA is copied, so that when the cell divides, both cells will have a copy of this genetic information. More precisely, at the beginning of the S phase, each chromosome is composed of one coiled DNA double helix molecule, which is called a chromatid. At the end of this phase each chromosome has two identical DNA double helix molecules, and therefore is composed of two sister chromatids. The end result is the existence of duplicated genetic material in the cell, which will eventually be divided into two. The S phase is followed by the G2 phase. [4] Figure 11 Cell Cycle The G2 phase occurs just before the cell begins to divide into two cells, and is a preparation stage for its chromosome duplication. The cell can then enter the M phase where the cell division occurs and two new cells are formed. The M phase is the mitosis phase and is divided into four subphases: prophase, metaphase, anaphase, and telophase. The term interphase refers to the period of cell growth that occurs before mitosis or cell division. Interphase occurs before mitosis or cell division and is composed of the phases G1, S, and G2 (Fig.11). During the interphase chromosomes are not visible, and DNA can be seen through a microscope only when a stain is used to make it visible. During the interphase the cell begins processes needed during mitosis or cell division. After cell division occurs, the cell may enter the resting G 0 phase. Although, it is not considered part of the cell 19

20 cycle, it may occur. This G 0 phase is very important when modeling radiation treatment of cancer because the cell is less sensitive to radiation when it is in this phase. [5] For mammalian cells growing in culture the S phase is usually in the range of 6-8 hours, M less than an hour, G2 in the range of 2-4 hours, and G1 from 1-8 hours, making the total cell cycle in the order of hours. In contrast, the cell cycle for stem cells in certain tissues is up to about 10 days. The radiosensitivity of cells varies considerably as they pass through the cell cycle. It seems to be a general tendency for cells in the S phase to be the most resistant and for cells in G2 and mitosis (M) to be the most sensitive. The reason for the resistance in S phase is though to be due to homologous recombination, increased as a result of the greater availability of undamaged sister template the S phase, so a sister chromatid is being used as a template to faithfully recreate the damaged section and join the ends together properly. Sensitivity in G2 probably results from the fact that those cells have little time to repair radiation damage before the cell is called upon to divide. In general, cells are most radiosensitive in the M and G2 phases, and most resistant in the late S phase.the cell cycle time of malignant cells is shorter than that of some normal tissue cells, but during regeneration after injury normal cells can proliferate faster.cell death for non-proliferating (static) cells is defined as the loss of a specific function, while for stem cells it is defined as the loss of reproductive integrity (reproductive death), A surviving cell that maintains its reproductive integrity and proliferates indefinitely is said to be clonogenic DNA Structure Deoxyribonucleic acid (DNA) is a large polymeric molecule that has a characteristic double-helix structure consisting of two strands, each made up of nucleotide building blocks. A nucleotide is a subunit in which a base is linked through a sugar group to a phosphate group. The sugar is deoxyribose, which has a five-atom ring: four carbons and one oxygen. The backbone of the molecule consists of alternating sugar-phosphate groups. There are four different bases. Two are single-ring groups (pyrimidines): Thymine and Cytosine. Two are double-ring groups (purines): Adenine and Guanine. It is the order of these bases along the molecule that specifies the genetic code. 20

21 The two strands of the double helix are held together by hydrogen bonding between the bases. These bonds are formed between thymine and adenine, and between cytosine and guanine; the bases are paired in this way along the length of the DNA molecule(fig.12). During the S phase of the cell cycle, DNA synthesis takes place (the process of replication) in which every base pair is accurately duplicated. [2] Figure 12 DNA structure- double strand The sequence of these bases determines the genetic code. The two chains of the double DNA strand are held together by hydrogen bonds that grow up between the above nitrogenous bases. Thymine is always connected with adenine (T-A) and cytosine is always connected with guanine (C-G). During the S phase, DNA synthesis occurs and the cell doubles its genetic material. Cell functions are achieved by proteins that are composed by the cell based upon the genetic code. The overall process includes the transcription of DNA into RNA. Transcription is performed by RNA polymerases, which bind to DNA and generate the corresponding messenger-rna (mrna). The control of transcription is increasingly being elucidated and involves specific DNA sequences at the beginning and ends of genes that signal the initiation and the end of transcription. RNA molecule has a similar structure with DNA molecule; the only difference is that the RNA molecule has ribose rather than deoxyribose and uracil (U) instead of thymine (T). The decryption of DNA is based upon the correspondence of A-U and C-G. 21

22 Figure 13 Coding of genetic information Protein synthesis takes place at the ribosomes. There another molecule of RNA, called transfer-rna (trna), with an amino acid attached (structural unit of proteins), recognises specific base sequences in the m-rna (codons)(fig.13). The amino acids transferred by this way are lined up and held together constituting a protein. [5] In the cell nucleus, the very long DNA double-helix molecule, together with nuclear proteins, is organized within the cell through a number of levels of supercoiling. The DNA double helix (about 2.5 nm in diameter)is first coiled around protein cores made of histone proteins to form a bead-like string of nucleosomes, then coiled into a 25 nm fibre, which is further spiralized and becomes visible in the condensed form of a chromosome at mitosis. During interphase and in chromatin that has been gently extracted from the cell nuclei, DNA shows a series of loops or domains that are attached to the nuclear matrix. A human chromosome may have around 2600 looped domains, each formed from about 0.4 mm of the 25 nm DNA fibre and containing base pairs. [2] DNA Damage DNA is the most important part of a cell that can be affected by radiation because it carries the genetic code. The most radiosensitive DNA component is the Pyrimidine bases. Other important cell molecules that can be affected by radiation are the enzymes and the proteins of cellular membranes. The peptide bond in 22

23 proteins is the most radiosensitive.radiation induces many kinds of lesions to the DNA molecule, many of which are repaired by the cell, others are transmitted to daughter cells, and a relatively small portion of them, leads to cellular death. The most significant radiation-induced DNA lesions per Gy of radiation are listed on the following table (Table 3). Type of damage Number per Gy Double strand brake (DSB) 50 Single strand brake(ssb) Base decay Sugar decay Cross-conjunction DNA-DNA 30 Cross-conjunction DNA-proteins 150 Table 3 Number and type of DNA radiation damages Early experiments showed that irradiation leads to a loss of viscosity in DNA solutions. Subsequently this has been shown to result from DNA strand breaks. There are two categories of DNA strand breaks; single-strand (SSB) and doublestrand breaks (DSB). The detection of these depends on a study of the size distribution of fragments of DNA after extraction from irradiated cells. [2] The DNA double strand brake (DSB) is therefore though to be the most important type of cellular damage. Just one residual DSB may be sufficient to produce a significant chromosome aberration and thus to sterilize the cell. However, since cells have the ability to repair some of those damages, not all the DSB give rise to cell death. According to Ward, DSB damage is lethal when they exist together with other DSB damages, but with SSB damages too. [15] Apart from certain kinds of cells (lymph cells, progamets and serous cells) that are destroyed during the S phase when irradiated, the rest mammalian cells undergo mitotic death, meaning that the cells do not immediately die, but cell death occurs after the next or some following divisions. We consider that a cell is functional if it can produce 50 new cells, that is to complete 5 to 6 mitosis (2 5 =32, 2 6 =54). This is explained by the fact that some RNA quantities are not affected by DNA lesions and can still compose normal protein, keeping on cellular functioning for a while. Chromosomal alterations that can be induced by ionizing radiation through the aforementioned ways are the following (Fig.14): 23

24 (i) (ii) (iii) Genetic mutations. These are alterations in the genetic code that can cause either cell death or the altered genetic code can be transmitted to the posterities. Quantitative alterations of the cellular DNA and creation of polypeptide gigantic cells. Chromosomal morphological alterations. These alterations are visible in irradiated cells during the mitosis phase. Irradiation causes fractures in different parts of the chromosome. The occurring fragments, tend to adhere to other parts of defective chromosomes, and not to whole ones resulting to chromosomes with uncommon morphology. Generally the following can happen: The fragments may reunify in the correct vacancies of the chromosome, causing the damage to be repaired. The fragments do no reunify and the chromosome remains with an eliminating part. The fragments reunify in various combinations and chromosomes with structural abnormalities occur. X-ray X-ray Chromosome at G1 phase Chromatid break Aberration replicated in mitosis Chromosome at G1 phase Two chromatids brakes Damage ends reform Aberrations replicated in mitosis X-ray X-ray X-ray. Two chromosomes in G1 phase Chromatid brake in each chromosome Chromosom es join Dicentric chromosome in mitosis Two differet chromosomes in G1 phase Chromatid fragments Fragments reform on wrong chromosomes Figure 14 Chromosomal alterations due to ionizing radiation 24

25 Cell death is induced by the above impairments in the net or in the few following cell divisions, and these are called unstable lesions. On the contrary, others that show milder chromosomal alterations are called stable because they do not cause cell death, but they are more dangerous as they are passed on to daughter cells. One important parameter that must be mentioned concerning the phase in which a cell is irradiated, has to do with the fact that if the cell is irradiated earlier than the DNA synthesis, the lesions refer to the to the whole chromosome, cause the genetic code duplication has not yet started at this point. On the other hand, if the cell is irradiated later on the cell cycle, chromosomes are divided into two chromatides and it is possible lesions to occur in one of them. [4] The radiation-induced effects on cells can be classified according to the percentage of lesions or alterations that a cell undergoes (Table 4). No damage Preservation of normal reproductive ability No lethal damage Preservation of normal reproductive ability after a recovering period Potentially lethal damage Preservation of normal reproductive ability after a recovering period in resting conditions Mitosis delay Delay in G2 and S phase Sub-lethal damage Preservation of normal reproductive ability, slow growing rate Lethal damage Loss of reproductive ability Table 4 Radiation impact on cells As shown in the Table 4, in radiobiology, the cell survival correlates highly with the preservation or not of the cells reproductive ability. Doses that are used in Radiotherapy cause usually mitotic death. However, if the dose received in one fraction is a few hundred Gy s, cell death occurs earlier than mitosis. The mean lethal dose leading to loss of reproductive ability, for mammalian cells, is approximately 15 Gy, though 50% of lymph cells are led to death during interphase, 24 hours after the administration of 10 Gy. 25

26 No lethal damage: As we have seen previously, radiation causes damage at the molecular structure of DNA double strand. The damages are concerned with alteration in bases, single or double strand brake of DNA strands and peculiar cross-conjunction between DNA and proteins. The most crucial damage is DSB (double strand brake) and is lethal if it is not repaired. No lethal damage accumulates in cell and becomes lethal. Lethal Damage: cell survival after irradiation is closely related to the conservation of its reproductive ability. With dose of several hundreds Grays, lethal damage occurs before mitosis, during interphase (interphase death). At doses used clinically, the most common is mitotic death. The mean dose required for the loss of mammalian cell s reproductive ability, is 150 cgy. When a cell dyes, the first thing observed is chromatin condensing, nucleus perceived as having no nucleus. Before cytolysis, gigantic cells often occur provided mitotic procedure has been inhibited the rest of cell s functions keep on and so the size of the cell grows bigger. Sub-lethal damage: Is related to cells that preserve their reproductive ability but they suffer genetic lesions that they pass to posterity. Damage like this can cause deceleration of cells proliferation rate as to make them more radiosensitive in next radiation exposure. Potentially lethal damage: It is observed that cell survival is appeared to be higher when irradiated cells are in condition of hypoxia, nutrient deprivation, in high density cultures or the protein synthesis had been inhibited. Survival increment is due to repair of the damage which is called potentially lethal damage. This kind of damage is lethal only if there is not enough time for the cell to repair the lesion. Mitotic delay: Regardless of cell experiencing a lethal damage or not, a delay in cell cycle progress is observed. The duration of the cell cycle is a function of dose, dose rate, cell type and the phase of the cell cycle during irradiation. The delay is observed mainly in G2 and at the beginning of S phase.mitosis delay is a result of the inhibition of protein synthesis which are crucial for mitotic phase. These proteins are synthesized, at the end of G2 phase and after the completion of the synthesis the cell enters mitotic phase. Mitosis delay doesn t seem to affect the clinical result in dose range used in radiotherapy. 26

27 1.4 Radiation effects on tissues Early-Late Responding Tissues Whenever radiation therapy is given with curative intent there is a risk of serious damage to normal tissues. The risk increases with radiation dose, as does the probability of local tumour. Tumour control rate depends on the radiation tolerance of the unavoidably irradiated normal tissues. However, tolerance is a complex concept and can be defined, in experimental studies on clinic or on laboratory animals, in relation to a particular end-point such as 50% moist desquamation of skin, 5% pneumonitis in lung or 1% paralysis following spinal cord irradiation. The biologic effect of radiation on tissues lies in the cells that are responsible for tissue renewal. These cells are called target-cells. Their radiosensitivity and response to changes of radiation parameters, defines greatly the radiation impact on the tissue. Depending on whether the tissue renewal and function is performed by the same cellular populations, normal tissues are divided in Hierarchical (H- type) and Flexible (F-type). When tissue renewal and function is performed by different cellular populations, like in the epidermis, oral and intestinal epithelia, they are called H type. Tissues without a recognizable separation between the renewal and functional compartments, in which at least some of the functional cells also have the capacity for self renewal, are called F type. The liver, kidney, lung and cells of the Central Nervous System are examples of flexible organization. [4] The distinction between H and F type, called cell kinetics, is directly correlated with the appearance time of biological effects and defines the way of their response to radiation and their induction as radiobiological parameters. When H-type tissues are irradiated cell populations with the higher mitotic activity are harmed, specifically the cells to be divided. This means, that the maturefunctional cells, that are destroyed according to tissue -dependent rate, are not adequately replaced. Normally the life time of cells in H-type tissues are of some days or weeks. In clinical radiotherapy, their response increases severely during fractionated radiotherapy but start to dissolve few days or weeks after the end of treatment. These reactions are called Early reactions and H-type tissues are called Early responding tissues (Fig.15). Typical examples of early reactions include 27

28 radiation-induced mucositis, dermatitis and bone marrow depletion. The biological mechanism behind the healing of early reactions is that stem cells which survive radiotherapy or which immigrate from outside the radiation field proliferate and restore the tissue. Figure 15 Sketch of typical cell survival curves for (A) early responding tissues and (B) for responding tissues. late F-type tissues proliferate slowly, therefore their response is observed after a long period of time that may extent up to years after the irradiation. They do not disappear but often progress chronically. These tissues are called Late responding tissues. Late responses are observed three up to six months after radiotherapy has ended. Examples of late radiation-induced reactions include radiation pneumonitis, fibrosis, necrosis and alteration of vessels The role of time factor When referring to cell survival after fractionated radiotherapy, Elkind & Sutton in 1959 [17] have stated: The total dose required for a certain cell survival levelpercentage is expected to increase with the increase of dose fractionation. 28

29 Through the gradually increasing experimental data other factors that influence cell survival after irradiation were discovered, such as oxygenation, cell proliferation and redistribution during the cell cycle. These factors are known as the 5 R s of radiobiology: 1) Repair: Cells that undergo none lethal damage from irradiation can repair the lesions through enzymatic mechanisms, if the cell is not to be irradiated for the next few hours. This explains the appearance of a shoulder in cell survival curves, in low dose areas. 2) Repopulation: Normal cells destroyed by radiation, are replaced through homeostatic processes, from the available cellular pools. This is achieved by 3 ways: Time reduction of cell cycle Increase of growth fraction Reduction of cell loss factor 3) Redistribution: Cellular radiosensitivity differs during the cell cycle. It is higher in M-phase and in the limit G1-S, whereas relative lower during S- phase. During irradiation cells in the most radiosensitive phase are those to die. At the same time, it is observed that irradiated cells accumulate to premitotic phase G2, resulting in a synchronization of the remaining cells. If the next dose is administrated the moment that cells pass to the phase where their radiosensitivity is higher, then the maximum injury is accomplished. However, it is difficult to define the most efficient times for the following irradiations, appointing redistribution of uncertain clinical importance. 4) Reoxygenation: Oxygen existence in irradiated cells is of great importance. Fractionated radiotherapy allows gradually better oxygenation of tumour cells, since their nutritional needs are reduced, due to cell death. Reoxygenation increases cell s radiosensitivity as well as the reproductive and remedial ability of malignant cells. Great efforts for improving the oxygenation of tumour cells have been made, resulting though to none significant clinical benefits. 29

30 5) Radiosensitivity (intrinsic): is described by the initial slope of survival curves [1]. The main 3 reasons of increased resistance of cells and tissues to radiation are: Low intrinsic radiosensitivity High repopulation rate Hypoxia Note that two of these processes, repair and repopulation, will tend to make the tissue more resistant to a second dose of radiation; however, redistribution and reoxygenation tend to make it more sensitive. These five factors modify the response of a tissue to repeated doses of radiation and are responsible for the slope of an isoeffect curve. The fifth factor (intrinsic radiosensitivity) affects the height of the isoeffect curve. 1.5 Cell survival models Cell Survival Curves- LQ Model A cell survival curve describes the relationship between the surviving fraction of cells, i.e., the fraction of irradiated cells that maintain their reproductive integrity (clonogenic cells), and the absorbed dose. Cell survival as a function of radiation dose is graphically represented by plotting the surviving fraction on a logarithmic scale on the ordinate against dose on a linear scale on the abscissa. Cell surviving fractions are determined with in-vitro or invivo techniques. Examples of survival curves for irradiation of cells by densely (A) and sparsely (B) ionizing radiation beams are sketched in Figure

31 Figure 16 Sketch of typical cell survival curves for (A) high LET (densely ionizing) radiation and (B) low LET (sparsely ionizing) radiation. The type of radiation influences the shape of the cell survival curves. Densely ionizing radiations exhibit a cell survival curve that is almost an exponential function of dose, shown by almost a straight line on the log-linear plot. For sparsely ionizing radiation, on the other hand, the curves show an initial slope followed by a shoulder region and then become nearly straight at higher doses. Factors that make cells less radiosensitive are: Oxygen removal to hypoxic state The addition of chemical radical-scavengers, The use of low dose-rates or multi-fractionated irradiation and Cells synchronized in the late-s phase of the cell cycle. Several mathematical methods of varying degrees of complexity have been developed to define the shape of cell survival curves, all based on the concept of random nature of energy deposition by radiation. The linear-quadratic model (LQ) has gained wide acceptance in the radiation oncology community (Fig.18). It is being used for the design of almost all new fractionation and dose rate regimens such as hyperfractionation, accelerated fractionation, Hypofractionation, high dose rate and pulsed brachytherapy, as well as combined external beam and brachytherapy. It describes the cellsurvival curve assuming that there are two components to cell kill by radiation: 31

32 (1) where S(D) is the fraction of cells surviving a dose D, α is a constant describing the initial slope of the cell survival curve, and β is a smaller constant describing the quadratic component of cell killing. The α, β constants are characteristic for different types of tissue. There are two kinds of radiation damage: (i) Damage caused with a single hit, where DNA is damaged in two places (Single track events) (Fig.17) (ii) Damage caused by different hits, in two different DNA places that must interact in order for a lethal damage to occur (Twotrack events). In case of no interaction taking place, or just one of two places is damaged, the cell undergoes non-lethal lesions that may be repaired. Figure 17 Biological damage caused by single track events (A) and two track events (B). Damage referring to first type hit is linear dependent of dose and equal with exp (-αd), whereas for the second type hit equals exp (-βd). Consequently, in lower doses or in lower dose rates, first type damages dominate, whereas increasing the dose second type damage start to appear. [4] 32

33 Figure 18 The Linear Quadratic model. Cell surviving curve Biologic Effect of Parameters a, b Models that describe cell survival (e. g LQ model) after irradiation consist of 2 cell damage components: a linear part proportional to dose and a curvilinear one. In the LQ model, the linear part is described by α coefficient. The damage caused by a simple hit (Damage Type α) is dominant in very low doses and in low dose rates. This is why survival curves are rectilinear at dose rates 1-2 Gy/hour. With this dose rate damage from multiple hits (Damage Type β) is negligible. The linear component is not only the most significant damage at doses used in clinical practises but the one that defines cell s radiosensitivity. This is why α coefficient is considered to be a measure of intrinsic radiosensitivity. In Fig.18 it is evident that the survival curve slope is strongly affected from β coefficient, when α coefficient is small. When the slope of linear damage is substantial, β-component does not influence the total slope of the curve. The difference in slopes is defined as recovery coefficient, and for the LQ model is equal to: recovery coefficient-survival factor=exp (2βd 2 ). This formula shows that the magnitude of cell recovery increases with dose. As shown in Fig.19 if we compare the slopes for a certain Survival Factor (e.g SF=0.01) it is obvious that the difference 33

34 between the a and total curves, is due to the addition of β-component [6]. The β contribution is decreased as dose rate decreases. Therefore, as far as the LQ model is concerned the β-component value can be considered to be a measure of cell survival. Figure 19 Biological importance of parameters α, β Biological significance of a, β ratio Coefficients α, β appear through the α/β ratio in the equations calculating the Biological Effective Uniform Dose and this ratio is characteristic for different kind of tissues. Since the dimensions of the parameters are α: Gy -1 and β: Gy -1, the dimension of α/β ratio is Gy and corresponds to the dose at which the linear contribution to damage (αd on the logarithmic scale) equals the quadratic contribution (βd 2 ). The mathematical expression of the above statement is: e (-αd) = e (-βd2) αd = βd 2 d = α/β (2) where α and β values are calculated by the method of maximum likelihood analysis and the calculated ones are those that fit better to the observed clinical data. The biologic basis of the way that different tissues response to fractionated radiotherapy is of great clinical importance and when the value of α/β ratio is low then the corresponding survival curve has a bigger convexity, as shown in Fig

35 Figure 20 Surviving Fraction as a function of dose and dependence by the type of damage Fig.21 clearly shows the difference in response of late and early responding tissues. As dose increases damage in late responding tissues increases too. Figure 21 Comparison of survival curves between early and late responding tissues Late responding tissues are more sensitive to dose fractionation than early responding ones and tumors. Two are the major reasons for this: 35

36 a) Cells that undergo mitotic death with low doses are capable of completing more mitosis than cells that that suffer the same damage with higher doses. This is why, the first although they are radiobiological dead, maintain their functional ability and the functional ability of the tissue as well. Taking into consideration that the mitotic rate of late responding tissues is slow, this explains why these tissues show a smaller slope ( larger shoulder ) in the initial part in the dose-response curves. Thus early responding tissues have a higher value of α/β than late responding ones so their survival curve shows a bigger slope. b) Due to the low proliferation rate, late responding tissues have more time to repair a potentially lethal damage. This explains the shoulder existence in the late responding survival curve. The following Table (Table 5) shows α/β ratio values for different kind of tissues and organs according to the radiation induced damages. Tissue/ Organ End-point α/β (Gy) 95% conf. lim.(gy) Early Reactions Erythema 8.8 [6.9 ; 11.6] Skin Erythema 12.3 [1.8 ; 22.8] Mucositis 9.3 [5.8 ; 17.9] Oral mucosa Mucositis 15 [-15 ; 45] Mucositis ~8 Late Reactions Telangiectasia 2.8 [1.7 ; 3.8] Skin Telangiectasia 2.6 [2.2 ; 3.3] Telangiectasia 2.8 [-0.1 ; 8.1] Subcutis Fibrosis 1.7 [0.6 ; 2.6] Impaired shoulder Muscle movement 3.5 [0.7 ; 6.2] Nerve Brachial plexopathy < 3.5 N/A Optic neuropathy 1.6 [-7 ; 10] Spinal cord Myelopathy < 3.3 N/A Eye Corneal injury 2.9 [-4 ; 10] Bowel Stricture/ perforation 3.9 [2.5 ; 5.3] Lung Pneumonitis 4.0 [2.2 ; 5.8] Lung fibrosis 3.1 [-0.2 ; 8.5] Various late effects 3.5 [1.1 ; 5.9] Head/ Neck Various late effects 4.0 [3.3 ; 5.0] Oral cavity Various late effects 0.8 [-0.6 ; 2.5] Table 5 Characteristic values of α/β ratio for different types of tissues and organs 36

37 1.5.4 Radiosensitivity The basic principles of the Reproductive Death law for cells radiosensitivity, introduced by Bergonie and Tribondeau [11] defines that tissue radiosensitivity is considered to be: Dependent on the metabolic activity. Cells with high metabolic activity show high radiosensitivity. Dependent on the proliferation rate and tissue growth rate. As the rate increases, radiosensitivity increases too. Inversely dependent on cell variation-alteration. Radiosensitivity is high in cells that are less altered than those that show high alteration. Dependent on the cell age in the cell cycle. Younger cells are more radiosensitive. Tissue damage depends not only on the radiosensitivity of the parenchymatous cells but also on the damage of the connective tissue and vessels that leads to reproductive reduction of the parenchymatous cells, concluding in tissue hypoplasia and atrophy. In Table 6 different types of tissues and organs are categorized according to their radiosensitinity. Although the impact of radiation on different tissues has many similarities, radiosensitivity and tissue function vary in such a way that it is necessary to examine each tissue separately. When a tissue is irradiated three are the main facts that are observed in the induced damage: 1) Damage dependence on dose-time factor, that is from the administrated dose in correlation with the time period administrated 2) Damage dependence on the dose-volume factor, as the irradiated volume increases the induced damage is more intense 3) Dependence on the dose rate, cgy per unit time 37

38 Radiosensitivity Tissues/ Organs High Lymph glands, spleen, thymus gland Testis Ovary Bone Marrow Small Intestine Skin Medium Eye Vessels Bone development Low Liver, Kidneys, Lungs Central & Peripheral Nervous System Endocrine glands Striatal Muscles Myocardium Connective tissue Skeleton Table 6 Characteristic Radiosensitivity for different types of tissues Fractionation The optimal distribution of dose over time has been a major issue throughout the history of radiotherapy and important progress has been made in this area over the past few years. The clinical evaluation and implementation of modified fractionation schedules based on biological rationales is an important focus of translational research in radiation oncology. The relationships between total dose and fraction number for late-responding tissues, early responding tissues and tumours provide basic information required to optimize radiotherapy dose per fraction. Nowadays, it is an established scientific fact that the impact of radiation on tissues decreases when dose is administrated in fractions. Moreover, normal tissues recover faster and more effectively than neoplasms. There are still many answered questions, such as the repopulation and redistribution rate and mechanisms and the radiosensitivity of both, tumours and normal tissues. Recently there is a question about accelerated repopulation [22, 23, 25]. It seems to take place approximately 2-5 weeks after radiotherapy has started and it is expressed as an increasing tumour proliferation rate. If this is accurate, in order to 38

39 handle this effect dose per fraction should be increased and the total treatment time must be reduced as minimal as possible. But this may come in contrast, with the time needed from normal tissue to recover. 1.6 Radiobiological approach and models Radiobiological approach in Radiotherapy optimization The clinical outcome of a technique in radiotherapy is always attached with a certain grade of uncertainty, when we are focused on the probability of Tumour Control (TCP) and the Normal Tissue Complication Probability (NTCP). This partially has to do with the fact that to consecutive sessions of the same treatment may differ a lot, because the nature of the biological impact of radiation is probabilistic in a microscopic level. Moreover, the differences at the cellular level between the patients are generally unknown. So for these reasons, the expected treatment outcome is expressed as a certain response probability in the total effect of radiation. Consequently, the optimization of radiotherapy depends strictly on the data used, that correlate with each individual patient. For this reason it is important to insert radiobiological parameters that will describe the volume dependence and the response of normal tissue of the irradiated volume, as well as the dose fractionation and the relation dose-time. The technological evolution in Radiodiagnostics (CT, SPECT) and in the radiotherapy field (IMRT, Tomotherapy) gives the opportunity to approach new realistic data for the position and the distribution of sensitive functional subunits of the organs when irradiated, and more accurate data for tolerance doses in correlation with the irradiated volume. Furthermore, the 3D dose distribution calculation for each patient allows more accurate estimation of the treatment efficiency. In order to achieve a strong relationship between the treatment planning and the clinical outcome, it is important to use this great amount of information, through the verification provided by the mathematical models. Consequently, the further goal is the induction of radiobiological models in order to evaluate treatment plans and the estimation of biological parameters based 39

40 on clinical trials, so as to predict the radiation induced impact and optimize the initial treatment plan Importance of radiobiological approach in radiotherapy The normal distribution of dose in radiotherapy is estimated by the treatment planning system. This system receives as data the patient s anatomy through CT scans (or MRI scans), the number and the type of radiation beams (beams geometry, type, energy, etc). The result of these calculations is a three dimensional dose distribution in the area of interest. According to the national protocols, the uncertainty in dose calculation must not exceed 5%. However, when a patient is simulated in the computer with an equivalent tissue simulation, it is not clinically correct because the response of different organs in radiation depends on many factors that are not taken into account during the treatment planning process. Moreover, others uncertainties that can affect and alter the expected outcome in radiotherapy are: Internal organ movements Uncertainty in the assessment of the target-volume and critical organs. Uncertainty in patient position when placed in treatment position. Individual tissue response in dose distribution for each patient. The biologic approach depends on many factors: dependence of the irradiated organs on the irradiated volume, internal structural organization of functional subunits for normal tissues and clonogenic density for targets, hypoxic cells and especially on the dose fractionation. The need of radiobiological approach lies in the goal of radiotherapy optimization, which is to conform the treatment to the patient s characteristics. This means, when designing the treatment plan the aforementioned factors that affect the clinical outcome must be included, in order to predict complications and design more accurately the treatment plan. 40

41 1.6.3 Radiobiological models The basic principle behind the mathematical quantification of the outcome in radiotherapy (i.e radiobiological models) as introduced by Munro and Gilbert in 1961 is the following: The goal of radiotherapy of cancer is the effective damage of every potential malignant cell, in such degree that it cannot continue proliferating. Based on this principle and the random nature of radiation induced damage, a mathematical formula was derived, for tumour treatment, when the tumour consists of N identical cells. Tumour doses are constrained, since irradiation of critical normal organs is usually unavoidable. Great efforts are spent to make it possible to choose the best treatment plan from those that are under consideration, knowing their probable clinical outcome, namely the tumor control probability (TCP) and normal tissue complication probability (NTCP). Thus, mathematical models, which are based on Poisson statistics and on the LQ Model of cell killing, quantify the radiobiological response of normal and malignant tissues in radiotherapy. The radiobiological parameters of the examined cell-response and cell survival models (i.e α, β, γ, S) must be separately calculated for some normal and malignant tissues. Specifically, there are two levels that can be approached mathematically: a) Microscopic level, studying cell survival b) Macroscopic level, studying organ response Exportation of dose-response relationships is dependent upon clinical data (treatment planning data and post-irradiation response assortment results) that are available for each patient. In general, these models predict a decreasing probability of tumor control accompanied with no side-effects on normal tissue, in relation to the increasing tumour volume and the increasing irradiated area of normal tissue. Some common characteristics of the models developed until now in order to describe the response of different normal tissues and tumours to radiation are presented below: Cell survival after irradiation is binomial and obeys to binomial or Poisson statistics. 41

42 The response of the entire organ depends on the death or survival of its celltargets (functional subunits for normal tissues and clonogenic cells for tumours). All the target cells respond similarly to radiation. If the intermediated time is enough, effects that obtained by similar dose distributions are considered to be the same too. The radiobiological model used more often in recent studies for the description of the dose-response distribution is the LQ Poisson model which takes into account dose fractionation of the treatment applied. Mathematically it is defined as: P(D)=exp[-N 0 e -(D/D50) (eγ-lnln2) ] = exp[-e eγ-and-βnd^2 ] (3) Where P(D) is the probability of tumour control or probability of normal tissue complication of an organ that is irradiated uniformly with a dose D, d is the dose per fraction and n is the number of fractions. D 50 is the dose that corresponds to 50% response and γ is the maximum normalized value of the slope of the dose-response curve. α and β are the parameters of the model that explain early and late effects on tissues. D 50 and γ are both dependent on the number of clonogenic cells of tumours and on the initial number of functional subunits for normal tissues. Parameters D 50 and γ (or α and β) receive specific values for each organ and for each endpoint and they can only be extracted by clinical data. A very important factor in radiobiological approach of a radiotherapy treatment is the way, through which post-irradiated complications of normal tissues are described by the models. This description is based upon Functional Subunits (FSU) inactivation. According to this theory, every organ consists of a set of FSUs which perform the function of the organ and have a specific structural organization. Structural organization of FSUs can be assorted in the following order: 1. Critical element 2. Integral response 3. Graded response 42

43 Specifically, critical element represents tissues of Serial FSU structure. This means that complication is emerged when at least one FSU is inactivated (tissues with serial structural organization are marrow and nerves). Integral and graded response represent tissues of Parallel FSU structure. In this case complication is emerged when a sufficient number of FSUs is inactivated. Moreover, another FSU structure had been proposed which is a combination of serial and parallel structure. Summing up, it is scientifically confirmed that structural organization of a tissue s FSUs, is of major importance in tissue response to radiation considering the dependence of that response by the volume irradiated. As the irradiated volume decreases, tolerance dose of the specific volume increases (for non-linear FSU structure). Normal tissue complication probability calculation is far more different than tumour control probability calculation. Calculation of normal tissue complication probability is strongly dependent on the accurate determination of their internal structural organization, however the structural organization of FSUs in tumours is considered to be absolutely parallel, so every single clonogenic tumour cell must be destroyed. Mathematical quantification of radiotherapy endpoint in the nature of radiobiological models which evaluate the probability of endpoint occurrence, are sectioned in two categories: 1) The first category comprises models that are referred to normal tissues having as subject of research the complication probability estimation (NTCP models). 2) The second category comprises models that refer to tumours; therefore the subject of research is the estimation of tumour control probability (TCP) Biological significance of the radiobiological parameters in the models. Linear-Quadratic model dispose parameters such as α and β which their biological significance had been thoroughly discussed in section 2.3 just as the biological significance of the α/β ratio. Steepness of the dose-response curve is another parameter that joins radiobiological models. Steepness is represented by the parameter γ in relative seriality model, by m in LKB model and by k in parallel model. A widely used method 43

44 to quantify the steepness of the dose-response curve is the normalized dose response gradient, or g value. This value defines the percent increase in response for a 1% increase in dose at a specified response level, usually 37% (g 37) or 50% (g 50) which are both in the steep part of the dose-response curve. Clinical dose response curves for TCP and NTCP are usually less steep than those calculated above for model situations. This is caused by heterogeneity, e.g. in tumour size, radiosensitivity of the tumour cells, repopulation during treatment, hypoxia or, for normal tissues, genetically determined radiosensitivity, co-existing disease of the irradiated tissues, pharmaceutical drugs (e.g. chemotherapy), or life-style habits such as smoking which may affect the radiation response. Even if we very carefully stratify patients for a clinical study into dose effect relationships, we will always have some (undetermined) heterogeneity in these and other factors. We therefore observe in clinical practice a composite dose-response curve of several underlying true doseresponse curves. The more heterogeneous the observed population is, the less steep are the dose-response curves. Usually dose response curves for normal tissue reactions, particularly late normal tissue reactions, are steeper than the doseresponse curves of tumours. This is easy to understand because important biological parameters in tumours are very heterogeneous, whereas normal tissues are expected to be much more similar between patients. TD 50 which enters LKB and parallel model, is always referred to the entire organ (or some reference volume) and like k, m or γ is a constant parameter that is characteristic of a given organ or tissue and a given endpoint. It is the dose which, if delivered uniformly to the entire organ, would result in a 50% probability of complication. Parameter n is a volume exponent, which depends on the organ and the endpoint. According to Laura A. Dawson et al, n represents the volume effect, which relates the tolerance doses of uniform whole organ irradiation to uniform partial organ irradiation. In other words, parameter n shows the sensitivity of the normal tissue complication probability (NTCP) to the irradiated volume. When n is near to 1, the volume effect is large and when it is near 0, the volume effect is small. [29]. Thus, higher values of n correspond to more resistant organs or parallel architecture (liver, lung) and low values of n correspond to more radiosensitive organ or serial architecture (bone marrow). s is the relative seriality parameter that characterizes the internal organization of the organ. A relative seriality close to zero (s 0) corresponds to a completely parallel structure, which becomes non-functional when all its functional subunits are damaged, whereas s 1 corresponds to a completely 44

45 serial structure which becomes non-functional when at least one functional subunit is damaged Dose-Response curves The goal of radiotherapy is to administrate such radiation dose to malignant cells in order to eliminate these cells, with minimized radiation effects in adjacent normal tissues. Clinical radiobiology is concerned with the relationship between a given physical absorbed dose and the resulting biological response, as well as the factors that influence this relationship. Quantification of biological effects due to radiation exposure, possess significant view in modern radiotherapy. Therefore, estimation of radiotherapy treatment result includes two parameters (biological results): Local tumour control and post-irradiation response of normal tissues. Disease local control is quantitatively referred as Tumor Control Probability (TCP) and response of normal tissue is quantitatively referred as Normal Tissue Complication Probability (NTCP) (Fig.23). Figure 23 TCP-NTCP relationship in clinical radiotherapy What is seen in clinical practice is a broad range of doses where the risk of a specific type of radiation reaction increases from 0% towards 100% with increasing dose, i.e a dose-response relationship. Radiation dose-response curves have a sigmoid shape, with the incidence of radiation effects tending to zero as dose tends to zero and tending to 100% at very 45

46 large doses. Many mathematical functions could be devised with these properties, one of the most common used ones is the Poisson. It is an empirical problem to decide whether one model fits observed data better than another. In reality, both clinical and experimental dose-response data are too noisy to allow statistical discrimination between these models, and in most cases they will give very similar fits to a data set. In this present study we used only the Poisson function Poison model Poisson distribution gives the probability of an event to occur specific number of times when the number of tests is high and the probability for the event to occur is very low. In order for the Poisson distribution to be implemented, event occurrences should be random and independent inter se. Poisson distribution is defined by a single variable, which is the mean number λ(x) of the event observations for the same time. P(x) = e -λ λ χ /χ! (4) Where P(x) is the probability for x observations and e is the napierian number (e=2.7). By definition 0! =1 and λ 0 =1. So the probability of zero observations is P(0)= e -λ. In the case of solid tumors irradiation, since damage of one cell by irradiation is random and independent from the damage of the rest cells, Poisson distribution is applied. The event studied in this case is cell survival and number of observations is represented by the variable x. If mean cell number that survive after irradiation of identical tumors is N cells per volume, then probability of tumor control is: P(x) = e -N N χ /χ! For 50% cure probability, that is x=0 and P (0) =0.5 we get: 0.5= e -N ln (0.5) = -N N= 0,693 This means that since 50% of the identical tumors consist of 0 clonogenic cells (cure), the rest 50% consists of 1, 2, 3 cells and these tumor cannot be locally controlled. Tumor numbers that contain survived clonogenic cells are distributed according to Poisson distribution. Percentages of tumors that contain 1,2,3 46

47 survived clonogenic cells, arise by placing x= 1,2,3 in turn. So for N= 0,693 we get P (1) = e /1! = 34.7% Similarly P (2)=12%, P(3)= 2.8% etc. So if 100 tumors are irradiated with such dose that 50 of them can be controlled, simultaneously 35 of them will contain 1 clonogenic cell, 12 tumors will contain 2 clonogenic cells etc. If it is assumed that N= N 0 e -kd (dose-survival relation) then TCP=exp (-N 0 e - kd N 0 ), where N 0 is the initial number of tumor clonogenic cells. According to this relationship, the characteristic sigmoid dose-response curve arises that it is concerned with local tumor control after irradiation and complications emergence in normal tissues. If dose-response relationship N= N 0 e -kd is been replaced by the Linear- Quadratic model (section 2.3) we come up with the equation that is most commonly used in clinical radiobiology and gives the probability of tumor control after fractionated radiotherapy of n fractions. TCP = exp [-N 0 exp(-αd-βdd+γτ)] Where d is the dose per fraction and D=nd is the total dose NTCP Models According to the level of mathematical approach, NTCP radiobiological models are sectioned in two categories: (i) Models that are based on microscopic response, and that is survival cell functioning: (a) Relative Seriality Model (b) Critical Element Model (c) Critical Volume Model (ii) Models that are based on macroscopic response of the organs: (d) LKB Model (Lyman, Kutcher and Burman Model) (e) Parallel Model 47

48 RELATIVE SERIALITY MODEL This model comprises a parameter (seriality, s) that describes functional behavior of a tissue. Thus response of an irradiated volume is dependent upon the combination of the parallel and serial structure of FSUs. Parallel and serial structure of FSUs was clearly expounded in section 2.6. Organs that consist of FSUs that are of serial organization have a low dependence by the volume irradiated, considering that every functional subunit is of vital importance for organ s function.organs that consist of FSUs that are of parallel organization have a high dependence by the volume irradiated, considering than a parallel organ can maintain its function even if a high portion of its FSUs have been destroyed. When parameter s (s for seriality) receives values that are very close to 0 (s 0) then this corresponds to a parallel organ such as lung or liver. However when parameter s receives values that are close to 1 (s 1) then the organ is considered to be serial with a very dependence on the volume irradiated. Parallel organs are bone marrow and esophagus. Thus normal tissue complication probability, P(D,V), for a uniform dose distribution is mathematically expressed as: P (D,V) = [1-(1-P(D,V ref ) s ) V/Vref ] 1/s (5) Whereas for a non-uniform dose distribution normal tissue complication probability is expressed as: P (D,V) = [1-Π(1- P(D i,v ref ) s ) vi ] 1/s (6) Where vi (= V i /V ref ) is a fraction of the differential irradiated volume ( V i ) of an organ divided by a reference volume V ref for which the parameters D 50 and γ have been calculated. P (D i, V ref ) is the probability of response that an organ of volume V ref develops when irradiated by a dose D i and is calculated by equation (3). The product Π parses from i=1 to M, where M is the number of the volume elements (voxels) that an organ has been theoretically splitted up. Usually the whole volume of the organ is considered as V ref since the organ as an aggregation is associated with body functional needs. Every organ inside the body is divided in volume elements (voxel), each of which has a definite volume of v i. During treatment planning process, every voxel yields a dose of D i and the 48

49 probability of response for every single voxel is calculated by equation (3). Then a value of the parameter s is determined according to the clinical outcome (e.g. radiation pneumonitis) after irradiation and response of the whole organ is calculated by equation (5) or (6) where all voxels are taken into account. LKB MODEL LKB model is widely used in calculation of normal tissue complication probability and was firstly proposed by Lyman in This model is based upon the assumption that the tolerance of a tissue while receiving homogenous radiation dose can be expressed by an exponential relation in which a parameter (volumetric exponent) n exists. Parameter n describes the dependency of the outcome by the irradiated volume. When n 1 then this dependency is high (lung), whereas when n 0 the dependency of the outcome by the irradiated volume is low (bone marrow). Thus, LKB model has 4 parameters that have to be calculated: V ref, D 50, m and n and is based upon ERF function (error function). Mathematically complication probability is described by the above equation: (7) Where, u(d,v) = [D D 50 (V/V ref )]/md 50 (V/V ref ) and D 50 (V/V ref ) = D 50 (1)(V/V ref ) -n, V ref is the reference volume for dose D 50, whereas V/V ref is the portion of the organ relative to V ref that is irradiated. D 50 (1) is the tolerance dose for 50% complication for a uniform irradiation of the whole organ and D 50 (V/V ref ) is the tolerance dose for 50% complication of a portion of the organ of volume V/V ref that is irradiated uniformly. The dependence of the probability of normal tissue complication by the irradiated volume is determined by the parameter n.slope of the dose-response curve using the LKB model is determined by the parameter m, which is conversely proportional with the parameter γ via the relationship: γ = π/8m. A serious constraint that must be taken into account by virtue of the presupposition that the dose is uniformly distributed is that in order for the LKB model to be used in clinical data, the non-uniform dose distribution has to be replaced by the corresponding equivalent uniform dose. 49

50 PARALLEL MODEL In parallel model the probability of normal tissue complication is a function of the ongoing number of FSUs that inactivated by radiation. Mathematically this is expressed by the following equation: P (D) = 1/[1+(D 50 /D) k ] (8) The above equation is a sigmoid dose-response function that describes the probability of damage of an FSU, again supposed a biologic equivalent uniform dose as discussed in LKB model. In this case the biologic equivalent uniform dose can be calculated using Linear-Quadratic (L-Q) formula. The parameters that enter the formula of the parallel model are D 50 which is the dose that causes 50% of the FSUs to get damaged and k which is the slope of the dose-response curve and determines the percentage by which the probability of a FSU damage increases with dose (k = 4γ). For a given dose-volume histogram (DVH) the fraction of the inactivated FSUs is calculated by: f= v i p (D i ) (9) Where v i and D i are the mean values of volume and dose obtained by the dosevolume histogram and f is the so called fractionated damage. 50

51 2. Methods and Materials PART I 2.1 Purpose of project Clinical application of different radiobiological models is provenly constraint not only due to difficulty in reliably describe complex radiobiological mechanism, but also due to lack of accurate knowledge of radiobiological parameters entering a model. A proper way of determining such parameters is by the use of clinical information as following up and record course of radiation therapy. Of course, this is an intent that is extremely demanding since it depends on many not wellunderstanding factors that are altered among different radiotherapeutic centers. Development of such models is based upon clinical information that is extracted by treatment planning. Such clinical data is a Dose-Volume Histogram (DVH) and clinical outcome of radiotherapy for each patient concerned. However, dose distribution by the form of a DVH in the region of interest, does not constitute the only crucial factor that describe biological effects caused by radiation. Also, a radiobiological model should comprise fractionation schedule of radiotherapy and the type of tissue response (early or late response). Information that concerns a DVH can be taken into account by the application of a Biological Equivalent Uniform Dose (BEUD). Notion of BEUD is going to be elaborately explained in later section. In the literature, dose-response relations and tolerance doses have been determined for different tissues and different clinical outcomes. However, many of these researches had based upon 2-D treatment planning and approximated determination of the clinical outcome in each time. So, accurate extraction of doseresponse relations by available clinical data leans on the accuracy of the derived clinical information. But, even when the accuracy of the available clinical data is of satisfactory level, radiobiological approach still remains a matter of challenge. This stands for the fact that available clinical data usually covers only a limited portion of dose-response curve (therapeutic range), so the portion of the curve that is out of the clinical dose range is based upon the form of the mathematical model used which confirmation is not possible at this range. The above limitation should be taken into account when data are to be used in other forms of radiotherapy either classic or 51

52 conformal (e.g. Intensity Modulated Radiation Therapy, which can cover a dose range similar to the range that is covered by the clinical data). The final issue of a radiobiological model is the determination of several parameters (e.g. D 50, γ and s for the relative seriality model) that join the models and are concerned with a specific tissue and a prescribed clinical outcome. These parameters determine the shape of the corresponding dose-response curves allowing the association of a certain dose distribution with the normal tissue complication or tumor control probability. Eventually, researchers haven t come up to a conclusion related to which parameters and models are more applicable and more accurate that can cover a wide range of clinical cases. Overall, radiobiological approach still remains a complex procedure even when clinical data is of great accuracy. Based on the above, the purpose of this study was focused on the following subjects: Theoretical approach: 1. Evaluation of the effect of different radiobiological parameters on the response of radiobiological models 2. Model comparison for different dose areas. Experimental approach: 1. Comparison of normal complication probabilities combined with different radiobiological parameters based on clinical data for breast cancer patients. 2. Estimation of Radiation Pneumonitis based on physical and biological parameters in lung cancer patients undergoing radiotherapy 2.2 Patient group- Method of treatment The research was based on 179 breast cancer patients treated with radiotherapy during the period of in the University Hospital of Tampere, in Finland. The patient group was selected without a conscious prejudice on the radiotherapeutic technique. 90 patients (51%) showed no symptoms and were used 52

53 as a control group. The rest of the patients showed various lung reactions starting from milder ones, such as cough, dyspnea during exercise or milder pain in the integumentary area of the breast (65 patients, 36%), and ending to severe symptoms such as severe cough and dyspnea during rest (24 patients, 13%). The total population was sub-divides in two groups: Ablated patients (43, 24%) and Resected patients (136, 76%). In order to define the treatment volume (PTV) and design the treatment plan, a CT scan was used. During the CT scan patients were placed in supine position, with their arms lifted and placed on their heads. PTV was defined in a 3D treatment planning system (CADPLAN), Varian. The same position was preserved during simulation and in the whole treatment process. The tomographic images were taken every 15 or 20 mm, covering the whole PTV area, with a pixel size of 1.3mm and slice thickness of 8mm. The treatment techniques included to opposed isocentric tangential photon fields (6MV) and in some cases an electron field was used for the thoracic wall and the parasternal and axillary lymph glands. The mean irradiated lung volume was 1333 ± 291 cm 3 (medial: 1452 cm 3, width: cm 3 ). The total dose was 50 Gy and was delivered during the period of 5 weeks. The normal dose distributions were corrected according to the normalized fractionation using the Linear-Quadratic Model with an α/β= 2.5 Gy. All doses in this study are corrected with the LQ model. The resected group was treated with to opposed isocentric tangential photon fields (6MV), targeting the breast, the thoracic wall and the peripheral lymph glands. Depending on the case, an extra field was used in order to treat the axillary lymph nodes and the superclavicularis. The ablated group was treated with to opposed photon fields and an electron field for the irradiation of the thoracic wall and the parasternal lymph nodes. The total dose was 50 Gy and delivered during the period of 5 weeks. The normal dose distributions were corrected according to the normalized fractionation using the Linear-Quadratic Model with an α/β= 2.5 Gy. All doses in this study are corrected according to the LQ model. 2.3 Dose Volume Histogramms 3D treatment planning available in radiotherapy allows for important information to be obtained such as detailed dose-distributions and resulting Dose Volume Histograms (DVH). DVH s provide very useful quantitative information as regards to the dose delivery in subvolumes of an organ. Data from DVH s enables 53

54 radiobiological models to be applied to assess the biological response, of healthy tissue, to radiation. Data from DVH s are used in this report for implementation and assessment of radiobiological models. Although useful, DVH data does come with many drawbacks. There is loss of spatial information on the dose-distribution, which can inaccurately present hotspots in regions of tissue where there is not such a pronounced problem. Parts of the Planning Target Volume (PTV) maybe evaluated as healthy tissue. [33]. Moreover a DVH does not differentiate between functionally or anatomical regions of an organ, making it difficult to identify structures such as FSU s [2] Calculations of NTCPs for each patient were performed using associated DVHs and the methods aforementioned in this report. A schematic of the calculation procedure which is neatly presented by Kwa et al. (1998) [18] is shown below: 3-D dose distribution DVH of an organ NTCP DVH reduction into a single parameter (BEUD, EUD) Figure 24 Calculation procedure of Normal Tissue Complicaion Probability An example of the received DVH data is shown in Fig.25. For each patient, the dose of each bin was provided followed by the associated volume. Calculations were performed using Microsoft Office Excel The available radiobiological models were developed in different worksheets. Normal tissue complication probability for every patient was developed in four different steps: 54

55 Complication probability P(Di) calculation for every lung volume element (voxel) Computation of the differential Volume ( ν i ) Calculation of overall normal tissue complication probability Plotting of different Dose-Response curves Percentage Volume Dose bins Patient ID Dose Volume Histogram_Resect ed Dose D corrected for fractionation Figure 25 Excel sheet with the Dose-Volume Histogram of each patient. After successful manipulation of the DVH data the procedure of NTCP calculation could be performed using each of the dose-response models mentioned in the next section of the investigation. Fig. 26, 27 show examples of the Excel environment in which the NTCP calculations were made: 55

56 Parameter set Complication Probability for each voxel Biological Effective Uniform Dose Figure 26 Excel sheet computing the probability for every dose bin 56

57 NTCP Model, parameter set Patient ID Dose- Response Curve Percentage Volume Differential Volume Complicati on Probability Figure 27 Excel sheet calculating the NTCP of each patient. Usually, the calculated radiobiological parameters describing the doseresponse relation of an organ refer to a certain uniform dose per fraction before deriving these parameters. Consequently, the application of the appropriate fractionation correction on the dose-volume histograms has to be seriously considered. In this study, the fractionation correction was applied using the linearquadratic model according to the formula above: (10) Where n is the number of fractions of the prescribed radiotherapy, D is the dose that every voxel receives, d is the daily dose given (1.8-2Gy) and the ratio α/β was elaborately depicted in section

58 Although this model is accurate for large doses it has not been validated for doses lower that 1Gy. Consequently, the correction may be approximative in this dose region, which is more relevant to normal tissues. [6] For the evaluation of a treatment plan, the mean dose of the dose distribution delivered to the tumor and its standard deviation are mainly used clinically. However, these data do not take into account the biological characteristics of the tumor. On the other hand, when different plans are compared to be classified one cannot also compare the effect of the treatment to the rest of the organs by using the mean dose to the tumor because the comparison does not use a common basis for all the plans as dose distributions throughout the target are never exactly uniform and may often be far from it. Solution to this problem was introduced by Niemierko et al. (1997) via the introduction of the Equivalent Uniform Dose (EUD) concept, which has since found applications in both external beam and brachytherapy. The EUD is defined as: The homogenous dose distribution which produces the same surviving fraction of clonogenic cells as that obtained with an inhomogenous dose distribution. [34] Another relevant concept of EUD is the Biologic Effective Uniform Dose (BEUD) in terms of tumour control or normal tissue complication. BEUD is the uniform dose that causes exactly the same total tumour control or normal tissue complication probability as a given non-uniform dose distribution on a complex patient case. This is based on the assumption that any two distributions are equivalent if they cause the same probability for tumour control or normal tissue complication. It is clear though that the concepts of EUD and BEUD have significant differences in their definitions and derivations; however, from philosophical point of view they both try to serve the same purpose. 2.4 Radiobiological Models The purpose of our investigation is to compare and assess the predictive strength of the most familiar normal tissue complication probability (NTCP) models, in predicting the incidence of Radiation pneumonitis following radiotherapy treatment. Three NTCP models are assessed; (i) the Relative Seriality model, (ii) the LKB model and (iii) the Parallel model, each with associated published parameter sets 58

59 (minimum parameter values, maximum parameter values) shown in Table 7. These parameters have in each case been obtained through clinical experiment. Min Max Mean RELATIVE SERIALITY D 50 = 15.1 Gy γ = 0.8 (Rancati & Gagliardi 2007) s= 0.04 D 50 = 34 Gy γ = 0.9 (Seppenwoolde 2003) s = 0.06 D 50 = 24.6 Gy γ = 0.85 (present study) s = 0.06 Min Max Mean LKB TD 50 = 15.3 Gy m = 0.28 (Rancati & Gagliardi 2007) n = 0.76 TD 50 = 30.8 Gy m = 0.37 (Seppenwoolde 2003) n = 0.99 TD 50 = 23.1 Gy m = 0.33 (present study) n = 0.88 PARALLEL TD 50 = 30.8 Gy m = 0.37 (Seppenwoolde 2003) n = 0.99 Table 7 Radiobiological parameter sets used in this study. 59

60 a) The Relative Seriality model. Firstly linear quadratic (LQ) model is modified as below: (11) Where P (D) indicates the probability of inducing the normal tissue complication (here radiation pneumonitis to lung) when it is irradiated uniformly with a dose D. The dose per fraction is d = D/n, where n is the number of fractions. D 50 is the dose, which gives a response probability of 50% and γ is the maximum normalized value of the dose-response gradient, which is located a little higher than the response point of 37% on the dose-response curve. The second equality of the above equation is valid in the region around D 50 and it gives the response probability using the second order approximation of a modified Poisson expression. As mentioned before each dose step in a volume-dose histogram was corrected for fraction according to LQ model using α/β = 2.5 Gy. Radiation sensitivity was assumed to be homogenous throughout the lung volume. Parameters D 50 and γ (or α and β) are specific for every organ and specific for the kind of injury (endpoint) considered and can only be derived from the clinical data. The complications observed in the normal tissues after radiation therapy have been described in terms of inactivation of functional subunits (FSUs). Many NTCP models try to account for the volume effect, which originates from the FSU infrastructure of the organs and describes how the tolerance dose increases with decreasing partial irradiated volume of the organ. The organization of the FSUs is described as serial, parallel or a combination of these two patterns. Organs with a parallel infrastructure have strong volume dependence since the organ can maintain most of its function even when a large proportion of its subunits is damaged. On the other hand, organs with a serial infrastructure have small volume dependence since every subunit is vital for organ function. The relative seriality model is one of the models that account for the volume effect. For a heterogeneous dose distribution, the response P I of normal tissues is given by the expression: (12) 60

61 Where P (Di) is the probability of response of an organ having the reference volume and being irradiated to dose Di as described by equation (11). vi = Vi/Vref is the fractional subvolume of the organ that is irradiated compared to the reference volume for which the values of D50 and γ were calculated. M is the total number of voxels or subvolumes in the organ, and s is the relative seriality parameter that characterizes the internal organization of the organ. A relative seriality close to zero (s 0) corresponds to a completely Parallel structure, which becomes non-functional when all its functional subunits are damaged, whereas s 1 corresponds to a completely Serial structure which becomes non-functional when at least one functional subunit is damaged. In this study, the whole lung constitutes the reference volume to which the model parameters D 50 and γ refer. The clinical validity of the relative seriality model, is proved by a number of published studies. [21] As discussed in previous section, due to the characteristic variability which exists in patient treatments, dose-distributions will be different. To allow a comparison on radiobiological terms between patients in a population and their subsequent dose-distributions, the concept of a biologically effective uniform dose BEUD is used. It is the dose that causes exactly the same tumour or Normal tissue complication probability in the lung as the original dose-distribution. [31] (13) This concept states that the uniform dose is biologically as effective as the dose delivered by an inhomogeneous distribution to the lung for each patient. [12] b) LKB and Parallel Model. In the LKB model, the NTCP estimations are based on a DVH reduction algorithm. Dose-volume histograms provide the dose distribution information of a non-uniform dose-delivery to a target volume. A multi-step DVH can be replaced by a single-step DVH with a uniform dose distribution. The DVH is reduced to a single value in dose-volume space and from this an NTCP can be calculated. 61

62 This corresponds to the transformation of an inhomogeneous dose distribution into an equivalent uniform irradiation of the whole lung with an isoeffective dose causing the same complication probability as the original DVH. The same NTCP induced by the inhomogeneous dose-distribution is caused by an equivalent uniform dose distribution EUD. The concept of EUD is useful when comparing dose response models and associated radiotherapy treatments. It is used in the same way as the aforementioned BEUD. For the LKB model, the dose distributions can be transformed into complication probabilities by using Lyman s four-parameter model, after the DVH has been reduced to the EUD with a power-law relationship such as: (14) In the same way, for the parallel or critical volume model, the mathematical expression of EUD changes to the following: (15) Both models are variations of the general parallel model, and in the special case that: D 50 or D 0 then EUDparallel = EUDLKB with n = 1/k. For cases of heterogeneously irradiated organs the following relationship for EUD has been proposed and used in LKB model: (16) In this formula differential-dvh (υ i = v i ) is used and Di is the dose corrected for fractionation for bin i. In order to calculate the normal tissue complication probability, NTCP the following mathematical formula has been used: 62

63 (17) Where t= (EUD-TD 50 )/m*td 50 Solution of the above equation (17) can be approached by the error function or erf that is computed as follows: A normalized form of normal distribution equation that gives the probability that a variable takes a value in interval [0, x] equals to: (18) And is related with the probability integral as: So, (19) (20) If we assume that u = t / 2 then du = dt/ 2 Then: (21) Where ERF is the error function and is easily calculated. Consequently, the probability of a variable to take a value between [x 1, x 2 ] interval can be calculated by the following formula: Φ(x) = ½ [erf(x 2 / 2) erf(x 1 / 2)] (22) 63

64 In the pointed case that someone refers to the interval [-, 0] then Φ(x) calculated by: (23) Especial attention must be provided during application of error function (ERF) in negative values of x: Φ(-x) = 1- Φ(x) = 1- Φ(-(-x)) (24) It is obvious from formula (17) that NTCP is determined by the dose that causes 50% of response (TD 50 ) and a slope parameter m (slope increases as m decreases). 2.5 Scoring Normal Tissue Complications The clinical outcome examined in this present study as a complication is Radiation Pneumonitis (RP). Radiation pneumonitis (RP) and pulmonary fibrosis (PF) represent acute and late phases in development of radiation-induced lung injury. Distinction between these phases is arbitrary because early and late effects of ionizing radiation on normal tissues constitute a continuous spectrum of biological events. For thoracic radiotherapy, RP and PF present the most common and the major, sometimes fatal, dose-limiting toxicities. Onset of RP occurs 1 to 6 months after treatment, whereas PF develops gradually months to years later. 64

65 Figure 28 Radiation Pneumonitis in X-ray and CT images Diagnosis of RP is based on non specific symptoms of dyspnea, cough, occasional fever, and chest pain with or without abnormalities in pulmonary function tests. Post treatment radiographic changes may reveal infiltration inside (local RP) or outside the irradiation field (diffuse RP), occasionally affecting the contralateral lung. In approximately 28% of RP patients, the diagnosis is uncertain due to confounding factors. PF develops in almost all patients receiving radiotherapy. It is detected radiographically through permanent scarring of the lung tissue. PF patients may have varying degrees of dyspnea, cough, and chest pain or present no symptoms. Although RP typically subsides over time, PF may progress and become irreversible. Radiation induced damages to the lung take place in three phases: First is the clinical manifestation of RP and is characterized by cell death with sloughing of type I pneumocytes and endothelial cells, release of surfactant, fibrin exudation in alveoli, decrease in macrophage counts, and occurrence of interstitial edema. The second step marks acute RP and is characterized by tissue reaction and inflammation with hyperplasia of type II pneumocytes; increase in leukocyte, macrophage, and fibroblast counts; obstruction of endothelia; and increase in collagen and elastin connective tissue fibers. Finally, PF shows generalized fibrosis with loss of capillaries, thickening of alveolar septa, and narrowing of alveoli. 65

66 There is no standardized approach in scoring Radiation Pneumonitis. The criteria considered significant-absence or presence of symptoms, required treatment, and radiographic changes in the lung-vary from one scoring system to another, making evaluation of RP ambiguous. Many scoring systems use the need for medical intervention to score RP. For instance, according to the Southwest Oncology Group, patients with grade 2 RP need steroids, whereas patients with grade 3 require oxygen. In contrast, according to Radiation Therapy Oncology Group scoring, patients with grade 3 RP require administration of both steroids and oxygen [36]. LENT-SOMA classification and the Common Toxicity Criteria version 210 include pulmonary function tests. All these scoring systems also use radiographic changes (pulmonary fibrosis) as part of their assessment. This variation in selection of scoring criteria limits accuracy of RP assessment as well as making risk probability modelling based on diverse clinical data problematic. Discrimination of patients at high and low risk of RP could facilitate dose escalation to the tumour at the same or lower level of normal tissue complication probability (NTCP) and allow inclusion of novel drugs and radiation modalities in the treatment schedules. [30] The evaluation of RP was retrospective, based on the clinical diagnosis and on the radiological findings. There is no standardized approach in scoring Radiation Pneumonitis. The criteria considered significant, absence or presence of symptoms, required treatment, and radiographic changes in the lung, vary from one scoring system to another, making evaluation of RP ambiguous. Different radiotherapy centers apply various classification systems in RP scoring, such as RTOG/ EORTC (Table Below), and LENT SOMA [27, 9, 37]. Complications are usually classified as binary results (yes/no) e. g RP, or as scaled results, depending on the severity of the complication, 0(none), 1 (mild), 2 (moderate), 3 (severe), 4 (severe, death). Binary results are correlated to the increased percentage of reaction to dose and not to the increasing intensity of the reaction. On the contrary, scaled results are combined from different facts and symptoms (such as density changes in CT exams and diarroia), which can be translated into scaled answers. In the case of breast cancer, probabilities of having radiation pneumonitis are very low. The reactions observed are mild and the traditional criteria might not be in able to depict reality. Consequently, the scoring criteria used, was a combination of the most important published toxicity criteria, RTOG/EORTC and LENT SOMA, so as to take into account the low percentage that RP holds as a clinical outcome. The symptoms that helped the clinical diagnosis (respiratory problems, pain in the chest area, etc) where re-evaluated and where decreased in a scoring system from 0-2, 66

67 instead of the 0-5 scale [21].In order to collect the clinical results, a protocol was used, containing questions about the symptoms a patient may have in the irradiated area as well as symptoms concerning the respiratory function (dyspnea, cough, fever, pain). Moreover, radiological findings in chest X-rays were examined and evaluated by radiologists, at the beginning of the therapy, 3, 6 and 12 months after therapy ended. A brief summary of the toxicity criteria used in this study are shown in the following Tables: Grade Pneumonitis Clinical Fibrosis 0 Minimal or mild symptoms of dry cough or dyspnea on Radiation evidence of radiation fibrosis extertion, without evidence of tumour progression or without or with minimal dyspnea. other etiology, with radiographic evidence of acute pneumonitis. 1 Persistent dry cough that requires narcotic antitussive Radiographic evidence of radiation fibrosis agents or steroid, or dyspnea with minimal effort but causing dyspnea with minimal effort but not at rest, without evidence of tumour progression or other at rest; does not interfere with activities etiology, with radiographic evidence of acute pneumonitis of daily living and requires steroid for treatment. 2 Severe cough, unresponsive to narcotic antitussive agent Radiographic evidence of radiation fibrosis or dyspnea at rest, with radiographic evidence of acute that causes dyspnea at rest, interferes pneumonitis, and requires oxygen (intermittent or with activities of daily living, and home continuous) for treatment. oxygen indicated 3 Radiation pneumonitis causes respiratory insufficiency, Radiation fibrosis causes respiratory requires assisted ventilation insufficiency, requires assisted ventilation 4 Radiation pneumonitis directly contributes to the cause Radiation fibrosis directly contributes to of the death the cause of death Table 8 RTOG/EORTC toxicity criteria SOMA (no RP) (Mild/no RP) (RP) (Severe RP) Clinical symptoms Mild symptoms/no Milder symptoms/ Intense symptoms/ Severe fibrosis or treatment milder treatment intense treatment pneumonitis Table 9 LENT/SOMA toxicity criteria 67

68 Grading Clinical symptoms No evident change or symptoms. Mild symptoms or dry cough or dyspnea during excerise or with least effort but not in rest. Pain in the breast area. Absense of severe symptoms: clinical and radiological evidence of pneumonitis, respiratory insufficiency, cough and dyspnea on rest. Table 10 Radiation Pneumonitis scoring system by Tsougkos et al. (2005) 2.6 Statistical Analysis Statistical analysis was performed to explore the existing relationships that exist between the NTCP values, predicted by each model, and the clinical outcome score of Radiation Pneumonitis. As Tsougos et al. investigated, there is a number of statistical methods that can be used in order to assess how well a model fits the clinical data. In this present study, Chi-square test and ROC Curves (Receiver Operating Characteristics) were used [21]. DVH data for each patient in the patient sample investigated was used to calculate the associated NTCP values, by applying the aforementioned models of dose-response, for each of the parameters shown in table (pinakas me set parametron modelon). Equivalent Uniform Dose (EUD) was incorporated into the analysis as an associated predictor variable, to examine the extent of its effect on the clinical outcome. The statistical models used in this study are briefly explained below: Chi-square test Chi-square test statistic is a statistical test that allows for a relationship between two categorical variables to be assessed. Categorical data is data that does not have a continuum. To look at the relationship between two categorical variables we analyze the frequencies, i.e. the number of cases that fall into each category. The chi-square statistic is a test statistic based on comparing the frequencies observed in certain categories, to frequencies you might expect to get in those categories by chance. Four important steps were performed to achieve the goal of statistical analysis: 68

69 1) Generate a hypothesis : [Radiation Pneumonitis is independent on dose] 2) Collect clinical data: [Clinical outcome-rp scores] 3) Fit a statistical model to the data: [This will test predictions] 4) Assess statistical model: [Does it support the initial predictions] Chi-squre is defined as: x 2 = sum of (((observed-expected number of individuals in cell) 2 )/expected). The sum is calculated by adding the results for all cells in a contingency table. The equivalent mathematical statement is: (25) in which O is the observed number of individuals (frequency) in a given cell, E is the expected number of individuals (frequency) in that cell, and the sum is over all the cells in the contingency table. The form of a contingency table is as shown below: OBSERVED No groups No RP(0) RP(1,2) Total Total Table 11 Contigency table of the observed data The observed data are the frequencies in the table, and to calculate the model data we calculate the expected values for each of the cells in the table using column and row totals: 69

70 (26) Where n is the number of observations, in our case 43. The contingency table of the observed data is modified as shown in Table below, giving the contingency table of the expected data. EXPECTED No groups No RP(0) RP(1,2) Total Total Table 12 Contigency table of the expected data Once this is done we enter the value for each of the cells in the table into Eq.25 and calculate the χ 2 value for each. We then sum the values to obtain a value for the χ 2 statistic. Like most test statistics, the distribution of x 2 depends on the number of treatments being compared. It also depends on the number of possible outcomes. To obtain the significance value it is important to know the degrees of freedom which are calculated through v=(r-1)*(c-1), where r is the number of rows, and c the number of columns. When analyzing v x v contingency table, the value of x 2 computed using the above formula and the theoretical x 2 distribution leads to P values that are smaller or bigger than they ought to be. If the observed frequencies are similar to the expected frequencies, x 2 will be a small number and if the observed and expected frequencies differ, x 2 will be a big number. Thus, the results are bigger or smaller concluding that our hypothesis is accepted or rejected. If x 2 decreases we are more confident that the experimental hypothesis is true [7].Our data were tested using a significance level of p 0.05 as well as p<0.01. In this investigation we have a scoring grade to assess the severity of irradiation effects and consequently radiation pneumonitis. The analysis was applied 70

71 for a single lung case. This scale is from 0-2 Tsougkos et al. (2005) and can be developed such as to produce a resulting dichotomous variable. In our analysis we have performed tests where patients with a score of 0 are considered not to have radiation pneumonitis, and patients with 1-2 are considered to have radiation pneumonitis, as well as tests were patients with 0-1 are considered not to have RP and patients with 2 are. Thus we have two categories in which patients can fall, (i) has radiation pneumonitis (ii) does not have radiation pneumonitis. What we also know from our data are the dose values BEUD/EUD for every patient. These values can be used to split the patients into different groups of ascending received dose. An important consideration for us is to ensure that the assumptions of the χ 2 test are met. The two that are important, are (i) that each patient contributes to only one cell of the contingency table i.e. we can t perform a repeated measures design [8], (ii) The expected frequencies should be greater than 5. Also an important point raised by Field [8] is that small differences in cell frequencies can result in statistically significant associations between variables. This makes it important to look at row and column percentages to interpret effects that we get, because percentages will not depend on sample size like frequencies do. A measure of the effect size in each case was made by calculating the odds ratio, a ratio that is a ratio of the probabilities of an event occurring and an event not occurring. Using the above mentioned categorical variables we can see if there is a relationship between the radiation pneumonitis score and the received doses. For example, from the contingency table giving the observed data (table above) we see that 19/43=44% do not have RP and 24/43=56% have radiation pneumonitis, with the assumption that score 0 corresponds to no RP, though score1-2 corresponds in RP. Based on our hypothesis that Radiation Pneumonitis is independent on dose, the percentage of RP appearance should be the same for each dose-group. This means that each group has 56% probability (6.14) to show Radiation Pneumonitis. Using the formula above we calculated x 2 for each case. By this way we have calculated the x 2 value for each patient group (ablated, resected and whole population) for every radiobiological model used in this study. 71

72 ROC curves (Receiver Operating Characteristics) A ROC curve allows us to investigate the predictive strength of different models, as well as to define the relationship between sensitivity and specificity.to evaluate ROC curves, x axis represents specificity (False Positive Rate, FPR) though y axis shows sensitivity (True Positive Rate, TPR). As a measure of distinction, we create a diagonal line in the graph, starting from point [0, 0] and ending to point [1,1]. As closer is the graph to the upper left part (point [0, 1]) so better is the test to distinguish between cases and no cases. A marker of the test adequacy is the region below the curve, as for an accurate test this region equals 1.0 though for a test that cannot differentiate is equal to 0.5 (it falls exactly on the diagonal line). When using any test, it is likely to make errors. Ideally, we would like to have values for specificity and sensitivity equal to 1. ROC calculations and curve designing where performed using Microsoft Office Excel We designed ROC curves for every patient group (ablated, resected and whole population) for every radiobiological model used in this study. Figure 29 Excel sheet computing ROC curves 72

73 PART II Protocol: Estimation of Radiation Pneumonitis based on physical and biological parameters in lung cancer patients undergoing radiotherapy. Introduction Radiotherapy in the chest area is commonly used for treatment of breast or lung cancer. Patients that show locally improving cancer or cannot be operated, undergo radiation therapy. Despite the technological improvements in radiotherapy, some techniques used in practice increase the absorbed dose locally, leading to a toxicity increase in the lung, which causes respiratory problems after radiotherapy has ended. Especially, when radiotherapy is combined with chemotherapy, this combination can increase even more the toxicity, a situation that can lead even to the patient s death. One of the most common radiation-induced injuries is Radiation Pneumonitis (RP), which is expressed as dyspnea, with cough with/without fever. This occurs due to the high radiosensitivity of the lung tissue, combined with the treatment plan that usually affects the neighbouring normal tissues. Radiation Pneumonitis Radiation pneumonitis (RP) and pulmonary fibrosis (PF) represent acute and late phases in development of radiation-induced lung injury. Distinction between these phases is arbitrary because early and late effects of ionizing radiation on normal tissues constitute a continuous spectrum of biological events. For thoracic radiotherapy, RP and PF present the most common and the major, sometimes fatal, dose-limiting toxicities. Onset of RP occurs 1 to 6 months after treatment, whereas PF develops gradually months to years later. Diagnosis of RP is based on non specific symptoms of dyspnea, cough, occasional fever, and chest pain with or without abnormalities in pulmonary function tests. Post treatment radiographic 73

74 changes may reveal infiltration inside (local RP) or outside the irradiation field (diffuse RP), occasionally affecting the contralateral lung. In approximately 28% of RP patients, the diagnosis is uncertain due to confounding factors. PF develops in almost all patients receiving radiotherapy. It is detected radiographically through permanent scarring of the lung tissue. PF patients may have varying degrees of dyspnea, cough, and chest pain or present no symptoms. Although RP typically subsides over time, PF may progress and become irreversible. Radiation induced damages to the lung take place in three phases: First is the clinical manifestation of RP and is characterized by cell death with sloughing of type I pneumocytes and endothelial cells, release of surfactant, fibrin exudation in alveoli, decrease in macrophage counts, and occurrence of interstitial edema. The second step marks acute RP and is characterized by tissue reaction and inflammation with hyperplasia of type II pneumocytes; increase in leukocyte, macrophage, and fibroblast counts; obstruction of endothelia; and increase in collagen and elastin connective tissue fibers. Finally, PF shows generalized fibrosis with loss of capillaries, thickening of alveolar septa, and narrowing of alveoli. 74

75 There is no standardized approach in scoring Radiation Pneumonitis. The criteria considered significant-absence or presence of symptoms, required treatment, and radiographic changes in the lung-vary from one scoring system to another, making evaluation of RP ambiguous. Many scoring systems use the need for medical intervention to score RP. For instance, according to the Southwest Oncology Group, patients with grade 2 RP need steroids, whereas patients with grade 3 require oxygen. In contrast, according to Radiation Therapy Oncology Group scoring, patients with grade 3 RP require administration of both steroids and oxygen. LENT- SOMA classification and the Common Toxicity Criteria version 210 include pulmonary function tests. All these scoring systems also use radiographic changes (pulmonary fibrosis) as part of their assessment. This variation in selection of scoring criteria limits accuracy of RP assessment as well as making risk probability modelling based on diverse clinical data problematic. Discrimination of patients at high and low risk of RP could facilitate dose escalation to the tumour at the same or lower level of normal tissue complication probability (NTCP) and allow inclusion of novel drugs and radiation modalities in the treatment schedules. RTOG Criteria for Radiation Pneumonitis Grade Pneumonitis Clinical Fibrosis 1 Minimal or mild symptoms of dry cough or dyspnea on Radiation evidence of radiation fibrosis extertion, without evidence of tumour progression or without or with minimal dyspnea. other etiology, with radiographic evidence of acute pneumonitis. 2 Persistent dry cough that requires narcotic antitussive Radiographic evidence of radiation fibrosis agents or steroid, or dyspnea with minimal effort but causing dyspnea with minimal effort but not at rest, without evidence of tumour progression or other at rest; does not interfere with activities etiology, with radiographic evidence of acute pneumonitis of daily living and requires steroid for treatment. 3 Severe cough, unresponsive to narcotic antitussive agent Radiographic evidence of radiation fibrosis or dyspnea at rest, with radiographic evidence of acute that causes dyspnea at rest, interferes pneumonitis, and requires oxygen (intermittent or with activities of daily living, and home continuous) for treatment. oxygen indicated 4 Radiation pneumonitis causes respiratory insufficiency, Radiation fibrosis causes respiratory requires assisted ventilation insufficiency, requires assisted ventilation 5 Radiation pneumonitis directly contributes to the cause Radiation fibrosis directly contributes to of the death the cause of death 75

76 SWOG (Southwestern Oncology Group) Toxicity Criteria Toxicity Grade 0 Grade 1 Grade 2 Grade 3 Grade 4 Pulmonary Normal Radiographic --- Changes with symptoms --- Fibrosis changes, no (also code symptoms symptoms) Toxicity Grade 0 Grade 1 Grade 2 Grade 3 Grade 4 Radiation Normal Radiographic changes, Steroids required or tap Oxygen Requires Pneumonitis symptoms do of effusion required assisted not require ventilation steroids It is often difficult to rank patients either in the radiation pneumonitis state or fibrosis state. The differences in laboratory findings as well as in the clinical exhibition of the patient are small, so we make an evident based hypothesis (conditional hypothesis) that: - patients that show progressing pulmonary insufficiency and cough, six (6) months after radiotherapy, and they respond to administrated steroids, are scored as radiation pneumonitis cases. - though, patients with progressive respiratory aggravation six months after radiotherapy, that do not respond to steroids, are scored as fibrosis cases. The appearance of RP is generally related to the radiation dose, to the fractionation scheme and to the irradiated lung volume. The evaluation of RP must be based on the clinical estimation, one to six months after radiotherapy has ended. In this study, the severity of RP will be scored according to the RTOG toxicity criteria. RTOG has published a reference table that uses grades from 0 to 5 to indicate the severity of RP, where 1 means that no symptoms are observed and 5 that the radiation effects led to the patient s death. This reference table is used internationally for scoring the specific side-effects of radiotherapy. According to the each case and mainly on the treatment position, the exposure of lung to radiation can differ between patients. Clinical experience has shown that the lung shows greater tolerance when partially irradiated rather than totally irradiated. Knowing the volume of the irradiated volume, is obviously very important when referring to Radiation Pneumonitis. In literature, there are data that refer to lung radiosensitivity, but in order to be representative it is necessary to know the 3-D dose distribution. 76

77 Using the 3-D treatment planning systems, that relate the dose/volume parameters quantitatively, we can estimate the Dose Volume Histograms (DVH s) for organs at risk, and consequently format mathematical models predicting the Normal Tissue Complication Probability (NTCP) from DVH s. However, the aforementioned models are based exclusively on dose-volume parameters but also on inaccurate estimations of organ radiosensitivity. Possible differences among the patients that refer to individual radiosensitivity or respiratory function are not taken into account. Moreover, another way to estimate the degree of lung radiation damage is Pulmonary Function Tests (PFTs). However, pulmonary fibrosis is a result of a variety of lung injuries. The damage on the endothelial and epithelial cells is the initial step leading to radiation pneumonitis. The healing process following irradiation is based on the activation of specific cells that compose oxidative stress products and important biological moderators, like cytokines. Studies made, indicate that observing some of these cytokines, may result in a useful tool to predict radiation pneumonitis as well as fibrosis. Defining these markers in the Exhaled Breath Condensate (EBC) is an appealing prospective of non-invasive estimation of the air-pore inflammation and of potential damage on the lung parenchymatous. Methods that have been used is to define markers such as the exhaled Nitrogen Monoxide (NO), the exhaled Carbon Monoxide (CO) and biological markers in the EBC, such as ph, Hydrogen Peroxide (H 2 O 2 ), Nitrics/Nitrites (NO 2 /NO 3 ), the 8- isoprostane, etc. None of the above markers have been studied up to now as a Radiation Pneumonitis indicator. The clinical estimation of radiation pneumonitis is often really difficult, since in most cases the differences between pneumonitis and infectious pneumonia or other pneumonological damages are almost untraceable. Therefore, there is a certain need of improving the clinical estimation of RP and this can be achieved by using High Resolution Computed Tomography (HRCT). Changes in lung density due to radiotherapy can be quantitatively estimated with findings in CT scans and correlation of these density differences with clinical symptoms of RP can be achieved. 77

78 Purpose The purpose of the present study is concentrated on the following topics-areas: Formation of a radiobiological predictive model for radiation induced complications and correlation with real clinical cases of RP, as well as expanding the correlation between the response of the pulmonary tissue and the absorbed dose. Correlation between the biological markers of PFT s with the irradiated lung volume. Investigating whether biological markers of the exhaled air and the blood plasma are correlated with potential damage of the pulmonary parenchymatous during cancer treatment. Optimize the clinical estimation of R.P, using quantitative estimations of lung density changes, following radiotherapy. Methods and Materials In this study, 50 patients with lung cancer that undergo radiotherapy are examined. These patients must fulfill the following criteria: Qualification Criteria: Entrance criteria Patients >18 years old that will undergo Radiotherapy with / without Chemotherapy Cytological/ Histological Diagnosis Stage: I-IIIB (without costal, pericardial accumulation) Normal hepatic and renal function PS:0-1 78

79 Exclusion criteria Another tumour Previous Radiotherapy in chest area Oxygen therapy Recent infection of respiratory system (1 month) Inhaled Steroids Chronic Obstructive Pulmonary Disease > stage ΙΙ, Pulmonary insufficiency, Oxygen therapy Former fibrotic damages in Thorax-CT (RP, Π, TBC) If brachytherapy is performed or surgical tumour ablation Hb>10gr/dl, WEU>1500, PLT> Hepatic or Renal insufficiency An Excel data base was created in order to collect patient s data, NTCP estimations and PFT results for each patient. Patients Data Name Gender Age Smoking Cancer Type Τ Ν Μ Radiotherapy Start End No. Sessions Total Dose Dose/ Fraction Irradiated Area Type of Radiotherapy Before Chemotherapy DATE EBC NO PFT's ph H2O2 NO2/NO3 VEGF FEV1 FVC DLCo 79

80 Before Radiotherapy DATE Radiological evidence EBC NO PFT's ph H2O2 NO2/NO3 VEGF FEV1 FVC DLCo Survey Protocol α) The radiobiological model used for the description of the response of the lung according to dose is the Poisson model: 2 eγ ( D/ D50 ) ( eγ ln ln 2) eγ αnd βnd PD ( ) = exp e = exp e where: P(D) is the probability of a certain complication to the lung (in this case Radiation Pneumonitis), when it is irradiated with dose D, and d=d/n is the dose per fraction with n being the number of sessions. D 50 is the dose that gives 50% probability response and γ is the maximum normalized slope value of the NTCP curve. The parameters D 50 and γ (or α and β) are specific for each organ and for the type of damage, so they can be extracted from the clinical data [28, 33]. NTCP P(D) 1 γ D Dose (Gy) 80

81 The number of lung or breast cancer patients being studied will be great enough (~50) in order to have good statistics and to extract the parameters, as well as the verification of the model on new patients. The clinical outcome will be scored as following: 1. Mild symptoms of dry cough or dypsnea during exercise 2. Persistant cough that requires antitussive compounds, dyspnea with minimum effort but no in relaxed state. 3. Serious cough that does not respond to antitussives, or dyspnea during relaxed state, clinical or radiodiagnostic evidence of acute pneumonitis, oxygen supply or steroids administration 4. Respiratory in sufficiency and continuous oxygen supply. b) For the evaluation of radiation-induced respiratory injury Pneumonological Function Tests (PFTs) will be used. Specifically, patients will undergo Spirometry and the Flow-Volume curve will be extracted. In this examination we are interested in the Forced Exhaled Volume in 1 sec (FEV1), in the Forced Vital Capacity (FVC) and in the FEV1/FVC ratio. 81

82 Moreover, the patient undergoes Diffusion where the lung diffusive ability on carbon monoxide is tested (DLco) (also static volumes of the lungs with the use of Helium (He) are examined). c) The exhaled Nitrogen Monoxide (NO) will be measured with a specific portable analyzer (NIOX MINO, Aerocrine, Sweden). 82

83 d) The Exhaled Breath Condensate (EBC) will be collected with a specific equipment (Ecoscreen, Viasys, USA) and for the definition of biological markers biochemical methods will be used, as it has been described in the past (Kostikas K, et al., Am J Respir Crit Care Med 2002), while the ph will be measured at the time of EBC collection with a certain electrode after deaeration with an inactive gas (Argon). The VEGF marker will me measured with an abzymic procedure (ELISA). The concentrations of the above cytokines will be evaluated by using a multivariation statistic model of correlation, as independent variables, in correlation with radiobiological and clinical parameters of the present study. e) Patients entering the protocol will be subjected to a CT scan before Radiotherapy, 1 1/2 and 4 months after radiotherapy has ended. So, comparison measurements for lung density before and after irradiation can be made, by studying X-rays made in certain times during the process of patient observation. High Resolution CT Normal Radiation Pneumonitis 83