Prognostication in Advanced Cancer: A Study Examining an Inflammation-Based Score

Transcription

1 Vol. 44 No. 2 August 2012 Journal of Pain and Symptom Management 161 Original Article Prognostication in Advanced Cancer: A Study Examining an Inflammation-Based Score Michael Partridge, MBChB, Marie Fallon, MBChB, MRCGP, FRCP, MD, Caroline Bray, MSc, Donald McMillan, PhD, Duncan Brown, MBChB, MD, FRCP, and Barry Laird, MBChB, MRCGP, MD, DRCOG, DFFP St. Columba s Hospice (M.P., D.B.), Edinburgh; Edinburgh Cancer Research Centre (M.F., B.L.), University of Edinburgh, Edinburgh; Clinical Trials Unit (C.B.), University of Glasgow, and The Beatson West of Scotland Cancer Centre (C.B.), Glasgow; and Department of Surgical Sciences (D.M.), University of Glasgow, and Glasgow Royal Infirmary (D.M.), Glasgow, United Kingdom Abstract Context. Prognostication in advanced cancer is challenging. Biomarkers of systemic inflammation (C-reactive protein and albumin) combined in the modified Glasgow Prognostic Score (mgps) have been used to assist prognostication in various cancer types. Objectives. The aim of this study was to examine whether an inflammation-based prognostic score (mgps) is useful in prognostication in advanced cancer patients. Methods. Cancer patients who had biomarkers (C-reactive protein and albumin) recorded were allocated an mgps ranging from 0e2. Groups were compared using Jonckheere-Terpstra and Chi-squared tests. Survival analyses were carried out using Kaplan-Meier and multivariate Cox regression models. Results. A total of 296 patients were included, and a representative subgroup of 102 had biomarkers recorded. The mgps was predictive of death (P ¼ 0.014) adjusted for sex, cancer site, age, hemoglobin, and white cell count. Patients with an mgps of 2 had 2.7 times the risk of death of those with an mgps of 0(P¼0.011). Patients with an mgps less than 2 had an 86.1% and 74.3% chance of being alive at two and four weeks, respectively. Conclusion. A role for the mgps in prognostication near the end of life is suggested. Biomarkers (e.g., mgps) may assist clinical decisions as to whether intensive treatments are appropriate and may facilitate end-of-life care planning. J Pain Symptom Manage 2012;44:161e167. Ó 2012 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved. Key Words Inflammation, prognosis, cancer, CRP, albumin, end-of-life care Drs. Brown and Laird are joint senior authors. Address correspondence to: Barry Laird, MBChB, MRCGP, MD, DRCOG, DFFP, Edinburgh Cancer Research Centre, University of Edinburgh, Western General Hospital, Crewe Road South, Edinburgh EH4 2XR, United Kingdom. blaird@ staffmail.ed.ac.uk or barry.laird@ed.ac.uk Accepted for publication: September 9, Ó 2012 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved. Introduction Prognostication in advanced cancer is a crude art form. It relies heavily on clinicians intuition/experience and draws largely from cancer trajectories. This is further compounded by the variability of an individual s clinical course /$ - see front matter doi: /j.jpainsymman

2 162 Partridge et al. Vol. 44 No. 2 August 2012 To illustrate, studies on survival prediction in advanced cancer patients have shown that clinicians are often inaccurate when estimating survival times. 2 One study found that only 20% of predictions were correct, with clinicians overestimating survival by a factor of up to five. 3 Accurate prognostication is needed, particularly as the intensive anticancer treatment (e.g., chemotherapy) of patients becomes more common near the end of life. The National Confidential Enquiry into Patient Outcome and Death examined patients who died within 30 days of receiving systemic anticancer therapy. 4 Eighty-six percent of these patients were receiving treatment with palliative intent and 35% of patients last cycles of chemotherapy were deemed to be inappropriate. The decision to give such treatments is often a difficult clinical judgment made withoutthesupportofscientificdata.inpatients nearing the end of life, accurate prognostication is of paramount importance. It may allow anticancer therapy to be given more appropriately, thus avoiding intensive anticancer treatments. It also allows patients and families an estimate of time remaining and may serve to facilitate end-of-life care planning. Prognostic tools have been developed for use in cancer, but a systematic review has highlighted that the majority of such tools have limited applicability. 5 Also, many studies in prognostication have focused on clinical markers (e.g., performance status) and neglected the increasingly important role that systemic inflammation may have. 5 Systemic inflammation is reported to be the seventh hallmark of cancer and is associated with cancer development and disease progression. 6 Inflammation in cancer may be a result of host-tumor interaction but also may be the result of a pre-existing proinflammatory state (the result of comorbidity or a genetic basis). 7 The systemic inflammatory response has been examined as a possible prognostic indicator in cancer. 8,9 The measurement of the systemic inflammatory response has been refined using C-reactive protein (CRP) and albumin. These two biomarkers have been combined in a prognostic scoring tool, the Glasgow Prognostic Score. The Glasgow Prognostic Score was subsequently refined to the modified Glasgow Prognostic Score (mgps). 10e13 The mgps has been studied extensively, culminating in a recent study of 10,000 cancer patients. 14 This study showed that an increasing mgps was predictive of reduced cancer-specific survival (independent of tumor stage) in all major cancer types. Although the mgps has been validated in the general cancer population, it has never been examined in cancer patients nearing the end of life. The primary aim of this study was to examine whether an inflammationbased prognostic score (mgps) is of use in cancer patients near the end of life. Secondary aims examine the usefulness of the mgps at two and four weeks, which are clinically significant time points. These aims were developed to examine the study hypothesis; the mgps is useful in prognostication in advanced cancer. Methods Study Design This was a retrospective analysis of patients admitted to a specialist palliative care center. The study was done in a 30-bed center (St. Columba s Hospice), which serves a population of around 350,000 in and around Edinburgh, U.K. This center was chosen because the population being studied was cancer patients near the end of life (most patients [60%] die during their admission). As this was an analysis of existing data, national ethical approval was not required; however, approval was given by the research committee of the institution. Patients All patients who were admitted between January 1 and October 31, 2006 and had a cancer diagnosis were eligible. The case records and laboratory results of all these patients were reviewed. Biomarkers Biomarker results of parameters that have been used in prognostication (CRP, albumin, hemoglobin, and white cell count) were examined. Results up to one week before admission, or the first blood test after admission that included CRP and albumin, were included. In patients with multiple admissions, data were taken from the first admission at which blood test results were available.

3 Vol. 44 No. 2 August 2012 Inflammatory Biomarkers in Prognostication 163 Calculation of mgps Patients with an elevated CRP (>10 mg/l) and hypoalbuminemia (<35 g/l) were allocated a score of 2. Patients with an elevated CRP (>10 mg/l) alone scored 1, and patients with a normal CRP (#10 mg/l) and any albumin concentration scored 0. These cutoffs are based on previous studies examining the mgps. 13,14 In previous studies, an increasing mgps has been associated with a poorer prognosis. Statistical Analysis The survival time, defined as the number of days from blood tests (CRP and albumin) until death, was calculated. At the time of the data analysis, those patients who remained alive were noted. As the average length of stay in a specialist palliative care center in the U.K. is 14 days, an analysis of survival at two weeks was carried out. An analysis of survival at four weeks also was carried out as it has been shown that inappropriate treatment is given to patients in the last four weeks of life. 4 All statistical analyses were performed in SPSS, version 15.0 (SPSS Inc., Chicago, IL). All statistical testing was conducted at the 5% level, so 95% confidence intervals (CIs) are reported throughout. The mgps groups were compared using a Jonckheere-Terpstra test for continuous variables and a Chi-squared test for trend for categorical variables. Survival analyses were carried out using Kaplan-Meier and multivariable Cox regression models. The assumption of proportional hazards necessary for the Cox model was met. Results During the study period, 296 patients with cancer were admitted. A subgroup of 102 patients had relevant biomarkers (CRP and albumin) recorded. There was no significant difference in sex (P ¼ 0.292), age (P ¼ 0.368), primary cancer site (P ¼ 0.934), hemoglobin (P ¼ 0.256), and white cell count (P ¼ 0.536) between the subgroup and the 194 patients in whom biomarkers were not available. There was no significant difference in survival time between the subgroup and those patients where biomarkers were not available, P ¼ (Cox regression model adjusted for age, sex, primary cancer site, hemoglobin, and white cell count). The subgroup of patients was, therefore, representative of the entire center population with respect to these factors. The relationship between mgps and patient characteristics is shown in Table 1.The median age was 72 years (interquartile range 60e78 years). Forty-six patients (45.1%) were male. Lung cancer was the most common tumor type, present in 31 (30.4%) patients. There was no difference in mgps between age, sex, or primary cancer site. Increasing mgps score was associated with lower hemoglobin (P ¼ 0.001). There was no relationship between white cell count and mgps. The relationship between mgps and survival is shown in Fig. 1. Overall survival decreased with increasing mgps. Eighty-nine patients died. The median follow-up time was 3.86 weeks. The results from the Cox regression model are shown in Table 2. After adjusting for sex, primary cancer site, age, hemoglobin, and white cell count, mgps remained predictive of death (P ¼ 0.014). Although those with an mgps of 1 were estimated to have 1.3 times the risk of death of those with an mgps of 0, this was not statistically significant (P ¼ 0.484; 95% CI 0.585e3.100). Those with an mgps of 2 were estimated to have 2.7 times the risk of death of those with an mgps of 0. This was statistically significant (P ¼ 0.011; 95% CI 1.252e5.875). Tables 3 and 4 show patient survival at two and four weeks (divided by mgps), respectively. Survival at Two Weeks Eighty-six percent of patients who had an mgps less than 2 were still alive at two weeks. The conclusion from these data was that there is at least a three in four chance (75%) that a patient who has an mgps less than 2 will still be alive at two weeks (P < 0.001). An mgps of 2 did not offer any advantage in determining whether patients were likely to be dead or alive at two weeks (P ¼ 0.056). Survival at Four Weeks Seventy-four percent of patients who had an mgps less than 2 were still alive at four weeks. The conclusion from these data is that there was at least a three in five chance (60%) that

4 164 Partridge et al. Vol. 44 No. 2 August 2012 Table 1 Relationship Between the mgps and Patient Characteristics mgps Characteristics mgps 0 (n ¼ 16) mgps 1 (n ¼ 20) mgps 2 (n ¼ 66) P-value Age, years, median (IQR) 73 (65e85) 76 (56e78) 71 (59e78) Sex, n (%) a Male 8 (17.4) 9 (19.6) 29 (63.0) Female 8 (14.3) 11 (19.6) 37 (66.1) Primary cancer site, n (%) a Lung 4 (12.9) 6 (19.4) 21 (67.7) Gastrointestinal 2 (8.3) 6 (25.0) 16 (66.7) Genitourinary 2 (11.1) 1 (5.6) 15 (83.3) Breast 3 (30.0) 3 (30.0) 4 (40.0) Other b 5 (26.3) 4 (21.1) 10 (52.6) Biomarkers, c median (IQR) Hb, g/l 127 (108e130) 122 (112e132) 109 (98e119) WCC (10 9 /L) 9.4 (6.1e12.1) 10.9 (9.4e13.2) 12.1 (8.6e17.4) mgps ¼ modified Glasgow Prognostic Score; IQR ¼ interquartile range; Hb ¼ hemoglobin; WCC ¼ white cell count. a Number of patients and % of group. b Other includes hematological, head and neck, and unknown primary cancers. c One patient in mgps 0 and four patients in mgps 2 did not have Hb or WCC recorded. a patient with an mgps less than 2 will still be alive at four weeks (P < 0.001). An mgps of 2 did not offer any advantage in determining whether patients were likely to be dead or alive at four weeks (P ¼ 0.117). Discussion The results of the study show the mgps to be predictive of survival in patients with cancer Fig. 1. The relationship between mgps and survival. The maximum survival times for mgps 0, 1, and 2 were 57.1, 30.9, and 36.0 weeks, respectively. near the end of life. As mgps increased, survival decreased. Patients with an mgps of 2 had a 2.7 times greater chance of dying than those with an mgps of 0. When specific time points are applied, the study suggests that an mgps less than 2 is useful in predicting whether a patient is likely to be in the last four weeks, or less, of life. The findings support that whereas an mgps less than 2 may be of use, an mgps of 2 does not offer any additional prognostic accuracy in the last four weeks of life. The results also demonstrate that the mgps was independent of tumor type, demonstrating its validity in isolation. It has previously been shown that systemic inflammation and cancer prognosis are linked and inflammation is a marker of prognosis, independent of tumor stage. 8,15,16 Previous studies evaluating the mgps have been in hospital-based patient groups where its role was in decision making regarding treatment. To our knowledge, this is first study in which biomarkers of systemic inflammation (mgps) have been examined as prognostic factors in cancer patients near the end of life. These findings may affect the care of patients with advanced cancer in several ways. Patients who have an mgps less than 2 are unlikely to die within four weeks and this may help inform management. Chemotherapy is often administered to patients near the end of life, and one study has suggested that up to

5 Vol. 44 No. 2 August 2012 Inflammatory Biomarkers in Prognostication 165 Table 2 Results from Cox Regression Model (Adjusted for Primary Cancer Site, Age, Sex, Hemoglobin, and White Cell Count) Relative Factor Hazard Ratio 95% CI P-value mgps mgps 0 1 mgps e mgps e Sex Male 1 Female e Primary cancer site Lung 1 Gastrointestinal e Breast e Genitourinary e Other malignancy e Age, years e Hemoglobin, g/l e White cell count (10 9 /L) e % of patients die within four weeks of chemotherapy. 17,18 It would seem inappropriate for chemotherapy to be administered so late in the disease; however, lack of accurate prognostic information makes this decision difficult. The mgps may assist in guiding treatment to those in whom it is appropriate. These findings may be of use in deciding on the appropriate place of care. Patients who are referred for end-of-life care with an mgps less than 2 are unlikely to die within four weeks. This may assist clinicians in deciding which patients are appropriate for admission to a specialist palliative care unit (which are generally not able to offer longer term care) and the Table 3 Relationship Between mgps and Patient Status at Two Weeks Status at Two Weeks Factor Dead Alive mgps < 2 n % (CI) 13.9 (2.6e25.2) 86.1 (74.8e97.4) mgps ¼ 2 n % (CI) 40.9 (29.0e52.8) 59.1 (47.2e71.0) n % (CI) 31.4 (22.4e40.4) 68.6 (59.6e77.6) Table 4 Relationship Between mgps and Patient Status at Four Weeks Status at Four Weeks Factor Dead Alive mgps < 2 n % (CI) 25.7 (11.2e40.2) 74.3 (59.8e88.8) mgps ¼ 2 n % (CI) 57.6 (45.7e69.5) 42.4 (30.5e54.3) n % (CI) 46.5 (36.8e56.3) 53.5 (43.7e63.2) Note: one patient was excluded from the analysis as he/she was lost to follow-up at 20 days; hence, status at four weeks was unknown. purpose of that admission, that is, symptom control vs. end-of-life care. Clinical prognostications are based on intuition, experience, and observation of the change in clinical condition, in particular, changes indicative of deterioration over a period of time. However, prognostication, particularly in the later stages, is difficult for all clinical staff, and this is an area where more accurate scoring systems may be of use. 1 If deterioration in clinical condition is not supported by biomarkers (e.g., an mgps less than 2), this may help guide treatment. An accurate knowledge of prognosis also may help facilitate discussion with patients and families. In the Western world at least, the concept of a good death is usually one free from distressing symptoms, in an environment in which one feels safe, valued, and at ease, occurring only once all goodbyes have been said and all affairs put in order. 19,20 More accurate prognostication may help facilitate discussion with patients and families about these issues, although it is not generally cited in studies as necessary in itself to achieve a good death. Thus, the true value of accurate prognostic information to patients, especially in the last weeks of life, remains unknown. Although the present study suggests that CRP and albumin may have a role in prognostication in advanced cancer, other biomarkers may be of use. White cell count and lactate dehydrogenase have been shown to be predictive of survival in advanced cancer. 21e23 Clinical factors (e.g., tumor type and performance

6 166 Partridge et al. Vol. 44 No. 2 August 2012 status) are also important and have previously been shown to be prognostic markers near the end of life in a large study of 1164 palliative care patients. 24 In prognostication, it has been suggested that the greatest accuracy is proposed to occur when a combination of subjective prognostic judgments and objective validated tools are used. 25 Future studies in prognostication, therefore, should combine factors that have been shown independently to be meaningful. 24,26 A reliable prognostic tool for use near the end of life, which draws together and systematically builds on parameters that have been shown to be useful (tumor stage, type, and performance status along with markers of systemic inflammation), would seem the ideal. Future studies also should be multicenter prospective studies where bias is limited. There are several limitations of this study. It was a small sample, single center, retrospective study at a specialist center, restricting the generalizability of the findings. Only a proportion of patients had biomarkers recorded, although this subgroup was representative of the whole population with respect to the factors recorded. Most patients had died during the study period; however, as a small proportion had not, the survival curves are censored. Also, the mgps as a prognostic marker in advanced cancer has not been validated in an independent validation set and this should be addressed in future work. In this study, only a single measurement was taken; however, in patients, the mgps may change over time. In real life, however, it would be reasonable to repeat mgps biomarkers, particularly when there were changes in the clinical condition. In addition, possible cofounders such as comorbidities were not taken into account and this has been shown to be an independent predictor of survival in other work. 27 Although prognostic factors are of use, these are different from predictive markers. The latter is a marker that predicts a differential efficacy of a particular therapy and could guide the choice of therapy. Prognostic information is probabilistic only; the prognosis of an individual shall always be either better or worse than the median of a group of patients at the same stage of the same disease. 28 It must be emphasized that the mgps is not accurate at prediction but does have some probabilistic value. Delivering a prognosis to a patient is, therefore, difficult, and there is a tendency to overestimate life expectancy. 2 The mgps may increase the accuracy of prognostication, which in turn may help improve patient care at the end of life. Conclusion The present study supports a possible role for the mgps in prognostication near the end of life. A key role of the cancer clinician is to inform patients and their families about treatments and assist them in decision making. 29 Biomarkers (e.g., mgps) may assist the clinical decision as to whether intensive treatments (palliative radiotherapy and/or chemotherapy) are appropriate in patients near the end of life. Biomarkers also may assist in guiding choice of the appropriate place of care for cancer patients, as specialist palliative care units are generally unable to offer longer term care. We would emphasize that biomarker-based prognostic information may complement the clinical impression but not dictate it, the principle of treating the patient, not the blood test being apt here. Furthermore, such biomarkers are just one important piece of the puzzle. Future studies should examine all proven bio- and clinical markers with the aim of developing a useful individualized prognostic tool for use in cancer patients nearing the end of life. Disclosures and Acknowledgments This work did not receive any specific funding. Dr. Barry Laird is supported by the European Palliative Care Research Center. Prof. Marie Fallon has received grants from Cancer Research U.K., Archimedes, Cephalon, and Pfizer Pharma. The authors declare no conflicts of interest. References 1. Oxenham D, Cornbleet MA. Accuracy of prediction of survival by different professional groups in a hospice. Palliat Med 1998;12:117e118.

7 Vol. 44 No. 2 August 2012 Inflammatory Biomarkers in Prognostication Glare P, Virik K, Jones M, et al. A systematic review of physicians survival predictions in terminally ill cancer patients. BMJ 2003;327:195e Christakis NA, Lamont EB. Extent and determinants of error in doctors prognoses in terminally ill patients: prospective cohort study. BMJ 2000;320: 469e National Confidential Enquiry into Patient Outcome and Death (NCEPOD). For better, for worse? A review of the care of patents who died within 30 days of receiving systemic anti-cancer therapy. London, UK: NCEPOD, Lau F, Cloutier-Fisher D, Kuziemsky C, et al. A systematic review of prognostic tools for estimating survival time in palliative care. J Palliat Care 2007;23:93e Mantovani A. Cancer: inflaming metastasis. Nature 2009;457:36e Mantovani A, Allavena P, Sica A, Balkwill F. Cancer-related inflammation. Nature 2008;454: 436e McMillan DC. Systemic inflammation, nutritional status and survival in patients with cancer. Curr Opin Clin Nutr Metab Care 2009;12:223e Roxburgh CS, McMillan DC. Role of systemic inflammatory response in predicting survival in patients with primary operable cancer. Future Oncol 2010;6:149e Forrest LM, McMillan DC, McArdle CS, Angerson WJ, Dunlop DJ. Comparison of an inflammation-based prognostic score (GPS) with performance status (ECOG) in patients receiving platinum-based chemotherapy for inoperable nonsmall-cell lung cancer. Br J Cancer 2004;90: 1704e Crumley AB, McMillan DC, McKernan M, McDonald AC, Stuart RC. Evaluation of an inflammation-based prognostic score in patients with inoperable gastro-oesophageal cancer. Br J Cancer 2006;94:637e Roxburgh CS, Salmond JM, Horgan PG, Oien KA, McMillan DC. Comparison of the prognostic value of inflammation-based pathologic and biochemical criteria in patients undergoing potentially curative resection for colorectal cancer. Ann Surg 2009;249:788e Roxburgh CS, Wallace AM, Guthrie GK, Horgan PG, McMillan DC. Comparison of the prognostic value of tumour and patient related factors in patients undergoing potentially curative surgery for colon cancer. Colorectal Dis 2010;12:987e Proctor MJ, Morrison DS, Talwar D, et al. An inflammation-based prognostic score (mgps) predicts cancer survival independent of tumour site: a Glasgow Inflammation Outcome Study. Br J Cancer 2011;104:726e Wigmore SJ, Fearon KC, Maingay JP, Ross JA. Down-regulation of the acute-phase response in patients with pancreatic cancer cachexia receiving oral eicosapentaenoic acid is mediated via suppression of interleukin-6. Clin Sci (Lond) 1997;92:215e Falconer JS, Fearon KC, Ross JA, et al. Acutephase protein response and survival duration of patients with pancreatic cancer. Cancer 1995;75: 2077e Earle CC, Neville BA, Landrum MB, et al. Trends in the aggressiveness of cancer care near the end of life. J Clin Oncol 2004;22:315e O Brien ME, Borthwick A, Rigg A, et al. Mortality within 30 days of chemotherapy: a clinical governance benchmarking issue for oncology patients. Br J Cancer 2006;95:1632e Steinhauser KE, Christakis NA, Clipp EC, et al. Factors considered important at the end of life by patients, family, physicians, and other care providers. JAMA 2000;284:2476e Clark J. Freedom from unpleasant symptoms is essential for a good death. BMJ 2003;327: MacDonald N, Kasymjanova S, Dobson H, et al. Prognostic value of baseline inflammatory markers in inoperable non-small cell lung cancer (NSCLC). J Clin Oncol 2006;24: Gripp S, Moeller S, B olke E, et al. Survival prediction in terminally ill cancer patients by clinical estimates, laboratory tests, and self-rated anxiety and depression. J Clin Oncol 2007;25:3313e Suh SY, Ahn HY. Lactate dehydrogenase as a prognostic factor for survival time of terminally ill cancer patients: a preliminary study. Eur J Cancer 2007;43:1051e Martin L, Watanabe S, Fainsinger R, et al. Prognostic factors in advanced cancer patients: use of patient-generated subjective global assessment in survival prediction. J Clin Oncol 2010;28:4376e Glare P, Sinclair C, Downing M, et al. Predicting survival in patients with advanced disease. Eur J Cancer 2008;44:1146e Gagnon B, Abrahamowicz M, Xiao Y, et al. Flexible modeling improves assessment of prognostic value of C-reactive protein in advanced non-small cell lung cancer. Br J Cancer 2010;102:1113e Roxburgh CS, Platt JJ, Leitch EF, et al. Relationship between preoperative comorbidity, systemic inflammatory response, and survival in patients undergoing curative resection for colorectal cancer. Ann Surg Oncol 2011;18:997e Selawry O. The individual and the median. In: Stoll B, ed. Mind and cancer prognosis. Chichester, UK: Wiley, 1979:39e Smith TJ, Swisher K. Telling the truth about terminal cancer. JAMA 1998;279:1746e1748.