MR Detection of Brain Iron

Transcription

1 MR Detection of Brain Iron Lara. Thomas, 1 Orest B. Boyko, 1 ' 2 Doglas C. Anthony, 2 and Peter C. Brger 2 PURPOSE: To provide frther qantitative stdies concerning the relationship with age between regional brain iron and T2 shortening. METHODS: a) Qantitative T2 calclations of eight anatomic regions (red ncles, sbstantia nigra, dentate ncles, corps callosm, cadate, ptamen, temporal lobe white matter, and frontal lobe white matter) from T2-weighted spin-echo images were performed in 6 patients aged newborn to 35 years. b) Qantitative brain iron concentrations were obtained in six of the eight anatomic regions (red ncles, sbstantia nigra, dentate ncles, corps callosm, cada, and ptamen) sing 13 atopsied brains (newborn to 78 years). Brain tisse from these six regions was digested with.6 HCI-2.5% wt/ vol KMn 4 for 2 hors at 6 C. After centrifgation,.1 ml of an iron-chelating reagent (2 moi/l ascorbic acid, 5 moi/l ammonim acetate, 6.5 nmoi/l ferrozine, 13.1 mmoi/l neocprine) was added and the absorbance was measred at 562 nm/l and compared with a standard crve with ferric chloride. c) The in vivo iron concentrations in tisse that were obtained were reprodced in for test tbe phantom stdies with ferric ammonim slfate or ferros ammonim slfate dissolved in either deionized water or 5% agarose. T2 calclations of the phantoms were made with a single-section mltiple repetition time, mltiple echo time acqisition. RESULTS: a) Clinical T2 calclationsall eight anatomic regions showed a decrease with age in T2 vale, beginning shortly after birth. Dring the first three decades, the T2 shortening was most significant in the region of sbstantia nigra. b) Qantitative brain iron-five anatomic regions bt not the corps callosm demonstrated an age-related increase in brain iron ( nmol/ g for the red ncles verss nmol/g for the corps callosm). c) T2 effect of iron in vitro--both the ferric and ferros iron phantoms showed a decreased T2 vale in the in vivo concentration range of iron obtained from the postmortem stdies. The T2 shortening was most marked for the ferric phantoms. COCLUSIO: There is an age-related accmlation of iron in five regions of the brain, correlating with an associated decrease in T2 vale that can be demonstrated in iron phantoms. Brain iron appears to contribte to the progressive decrease of T2 signal that occrs with aging. Index terms: Iron, brain; Brain, magnetic resonance; Age and aging AJR 14: , Sep/Oct 1993 It is clear that the signal intensity on T2- weighted images of cerebral tisses shows considerable variability in regard to both regional anatomy and patient age (1). Stdies correlating T2 signal shortening with the semiqantitative estimation of brain iron as inferred from wholebrain sections stained by the Perl method conclded that the deposition of this mineral was Received December 26, 1991 ; revision reqested March 11, 1992, received October 7, and accepted October 21. Departments of Radiology' and Pathology>, Dke University Medical Center, Drham, C Address reprint reqests to Peter C. Brger, MD. AJR 14: , Sep/ Oct / 93/ American Society of eroradiology important for the mechanism of T2 shortening (2-4). Sbseqent stdies have challenged this conclsion (5, 6), one investigation sggesting that T2 signal intensity and the concentrations of iron are not closely related (6). This stdy was ndertaken in an attempt to investigate this isse by determining: a) the in vivo regional change in T2 vale with age on MR scans of healthy patients; b) the iron concentration in these same anatomic regions as a fnction of age in postmortem stdies; and c) the concentration of iron that was determined in the postmortem stdies and whether it is of sfficient magnitde to alter the T2 signal in aqeos soltion and agarose gels. 143

2 144 THOMAS AJR: 14, September /October 1993 A B c Fig. 1. Progressive iron deposition with age in the R and S with T2-weighted spinecho images. A, In a 7-month-old girl, there is sbtle T2 shortening in the R (arrow) and region of the S (arrowhead) (28/8) relative to the bright signal of the white matter of the temporal lobes (open arrowhead). 8, A 4-year-old patient demonstrates evolving T2 shortening in the R (25/8). ote the T2 shortening de to myelination of the white matter in the temporal lobe. C, A 17-year-old patient demonstrates frther conspicos T2 shortening in the R and S (28/8). These areas are isointense to areas of T2 shortening, sch as the temporal lobe white matter and the anterior commissre (arrows). D, A 48-year-old patient demonstrates frther dark signal in the R and S (28/ 8). D Methods Clinical scans MR proton images in 6 patients withot any mass lesions by spin-echo plse seqence were randomly selected. Ages ranged from newborn to 85 years. Patients were exclded from the stdy if there was a gross radiographic abnormality sch as tmor, hemorrhage, or postsrgical changes. Qantitative T2 calclations were made from operator-selected regions of interest with standard software (GE ::>igna Image Analysis fnctions) with an iterative x 2 minimization program. T2 vales were obtained by the se of the first and second echoes from a dal-echo acqisition. On three scans, images throgh the basal ganglia were not archived and T2 vales for the cadate (CA) or ptamen were not available. In pediatric brains, sing the standard adlt 2-cm field-of-view generated images where the crsor size was occasionally larger than the: region of interest (for example, the red ncles [R] and sbstantia nigra [S]), we centered the crsor over the region of interest to minimize overlap. Eight anatomical regions (R, S, dentate ncles [D], corps callosm, CA, PM, and bilaterally, temporal lobe white matter) and frontal lobe white matter were stdied. Initially, the globs pallids was stdied also, bt becase of the heterogenos anatomic organization with gray matter, the presence of white matter, and blood vessels with freqently enlarged perivasclar spaces, we cold not obtain intrapatient T2 vales that were as reprodcible as were those for other regions. The globs pallids was, therefore, not inclded in the regions stdied. Images were obtained with a 1.5-T GE Signa system, T2-weighted spin-echo plse seqence (27-32/3, 8/1 (repetition time/echo time/excitations) acqisition matrix 256 X 256) to prodce 5-mm sections with a 2.5-mm intersection gap in the axial plane. The data was plotted with graphics software (Cricket Graph; Cricket Software). Measrement of brain iron Samples from 13 formalin-fixed brains were obtained from the Atopsy Service of the Dke University Medical Center. Brains had been fixed in 2% formalin for 2 to 4 weeks. Brains were exclded from selection if there was any intracranial abnormality at the time of atopsy or if a nerologic disease had been noted dring life. Tisse samples,.1 to 1. g, from six anatomic regions (R, S, D, corps callosm, CA, ptamen) were homogenized in 1 ml of distilled water, and 1.-ml aliqots were digested (7) in triplicate with.6 HCI-2.25 % wt/vol KMn 4 for 2 hors at 6 C in a water bath. The samples were stabilized at room temperatre for 3 mintes and then centrifged at 12, X g for 5 mintes. After the addition of.1 ml of an iron-chelating reagent (2 moi/l ascorbic acid, 5 mol/ L ammonim acetate, 6.5 mmoi/l ferrozine, 13.1 mmol/

3 AJR: 14, September/October 1993 BRAI IRO , c 8 :; <J ;; A I Sbstantia nigra.' I I 6 '\ :r : I Patient age (yea rs } Dentate ncles c 8.:: ;; :; <.> ;; 6.l4j,..,..... I c Patient age (yea r s} 1 <J Ptamen I c 8.::!! a 'l' Patient age (years} E Sbstantia nigra ::>!:. :5- oo..,_ 1 1 ge.::..s g on Patient age (yea rs} c: o ::>.!::. c:- E o v:::: 1 g E.!::E. 2 1 Dentate ncles oo Patient age (yea rs} Ptamen o oco o o Patient age (yea r s} 1 Red ncles <J Red ncles g 2 c 8..2 E, :'\ =-.. B 1 E Patient age (years) Patient age (yea rs) - 1 <J., c: 8.2!! <J -;:; 6 D Cadate Patient age (yea rs) c o ::> c =.. E o'' v:::: 1 ge.!::5 Cadate oo o o <D Patient age (yea rs} Fig. 2. Reslts of clinical MR scans and iron measrement by patient age in gray matter. A, S. B, R. C, D. D, CA., ptamen. In each anatomic region, there is a progressive fall of T2 signal in the clinical scans by patient age (left panels). This decline is steepest dring the first two decades of life and more gradal thereafter. Measrement of brain iron in these same five regions (right panels) demonstrates an increase in brain iron by age. The crves are steepest throgh the end of the first two decades and show a more gradal rise thereafter. A 11, , 1, , Corps Callosm 1 Patient age (yea rs) c t = f:!... B 2 6 Patient agp (years} Corp s Callosm 2 4 so c Patient ag (years) 8 Fig. 3. Reslts of clinical MR scans and iron measrement by patient age in white matter. A, T2 vales in the temporal and frontal white matter of the right and left sides (x, r.ight temporal lobe white matter; e, left temporal lobe white matter;. right frontal lobe white matter;, left frontal lobe white matter). B, T2 vales in the corps callosm. C, Iron concentration measred in the corps callosm. Early decrease in T2 vales is seen coinciding with myelination of the central nervos system, occrring predominantly before the age of two. There is no detectable change in T2 beyond this age, and there is no detectable change in iron levels over the ages stdied. L neocprine), the absorbance was measred at 562 nm and compared with a standard crve of ferric chloride. T2 effect of iron in vitro To evalate the relationship between T2 shortening and iron concentrations in vitro, ferric ammonim slfate and ferros ammonim slfate soltions were serially dilted to prodce concentrations of,.2,.5,.75, 1.5, and 2. mmol/l in deionized water or in.5% agarose. Samples were then scanned in 1 -ml plastic tbes placed in a styrofoam container in the same MR scanner with the same qadratre head coil sed clinically. T2 images of the

4 146 THOMAS AJR: 14, September/October 1993 ' :2 4 g g 4 5 c: c:.:2 t:. ;; :; B 2 3 t:. t:.l!.t:. 1 2 t:. t:.t:. t:. ;:! f- ' t:. t: Iron conccntn1tion (rnm) Iron concentration (mm ) Fig. 4. Relationship between iron concentration (e. ferros ion; D., ferric ion) and T2 signal in iron-containing soltions (a) and in iron-containing agarose phantoms (b). An increase of ferros and ferric ion concentration within the physiologic range depresses the T2 signal. (In water, 1 mmol/ L corresponds to 1 nmol/g.) phantoms were acqired with a 15-mm single section, mlti-tr (32, 16, 8, 4, 2 milliseconds) and mlti-te (2, 4, 6, 8 milliseconds) spin-echo acqisition at 256 X 192, 1 signal averaged, 22-cm field of view. Qantitative T2 vales were obtained at three different image levels along the tbes, again with the GE software image analysis package, and all measrements were done in dplicate. The final plotted T2 calclation was the average of the six determinations. Reslts Clinical scans All eight anatomic regions showed a decrease with age in T2 vale on clinical MR scans (Figs. 1 throgh 3). This decrease began shortly after birth and contined throgh the first three decades in iron-containing gray-matter strctres or over the first 2 years in white matter. There was a sggestion of a gradal decrease in T2 throghot the sbseqent decades in iron-containing gray-matter strctres only. There was considerable overlap in some regions between patients. In all ages over 6 months, the T2 signal was shortest in the region of the S, followed in trn by the R, D, ptamen, and CA ncles. Measrement of brain iron All six anatomic regions, with the exception of the corps callosm, showed an increase in brain iron with age (Fig. 2). This increase was detected as early as 6 months and rose rapidly with a flattening of the crves thereafter. The mean amont of iron in patients over 4 (mean age, 67.5 years), was in nmol/g of tisse, for the R, 144. for the S, for the ptamen, 185. for the D, for theca ncles and for the corps callosm. The level of iron was lowest in the corps callosm and was independent of age (Fig. 3c). T2 effect of iron in vitro The in vitro effect on T2 vales of iron in phantoms in the concentration range obtained from the postmortem work was determined (Fig. 4). The effect of increasing the concentration of ferric and ferros ions on the T2 vale in aqeos soltions is shown in Figre 4a. The presence of ferric and ferros ions in soltion decreased the T2 nmber, with the T2 shortened by 5% at 2 mmol/l ferros ammonim slfate and by more than 9% at the same concentration of ferric ions. The effect was especially prononced for the ferric soltion. The relationship between iron concentration and T2 vale in agarose phantoms is shown in Figre 4b. Althogh the magnitde of T2 shortening is less than in the aqeos soltion, the presence of either ferric or ferros ions clearly demonstrated shortened T2 relaxation. Discssion The in vivo experiments presented here recorded a fall with age in the T2 vale in all five gray matter and three white matter strctres stdied, as has been noted daily in clinical practice and as has been docmented in previos systematic stdies (1). By stdying scans of living patients, we fond it possible to accmlate a large nmber of vales for T2 in nondiseased tisse. The T2 shortening began shortly after birth and contined dring life, althogh the majority of the change had occrred by the end of the third decade. Becase these same areas are those that are deeply and progressively stained by the Perl method (3, 8-14), it seems reasonable to conclde that the amont of iron and the T2 change are related. This relationship is spported by the well-recognized ability of MR to detect hemosiderin arond chronic hemorrhagic foci (15-17). If the progressive shortening of the T2 signal is related to local iron concentration, then one shold be able to docment a relationship between the magnitde of T2 shortening and the regional concentration of iron. This stdy records sch a relationship with the age-dependent accmlation of iron coinciding with the rate of change of T2 signal. Similar reslts were reported by Hallgren and Sorander (1, 1, 12, 18). Despite these concrrent changes in iron concentration and T2 signal, Chen et a! (6) have sggested that, althogh iron in pathologic states sch as arond hemorrhages can affect T2 relax-

5 AJR: 14, September/October 1993 BRAI IRO 147 ation time and shorten T2, the normal concentration of iron, even in the aged, is below the threshold of detection by MR. We therefore imaged iron-containing soltions and agarose phantoms to determine whether the presence of iron in the physiologic concentration range is associated with a change in T2 signal intensity and vale. In aqeos soltions, increasing the concentration of both ferros and ferric ions showed a profond effect on T2, an effect mch greater than the change in T2 signal with age in normal patients. The crves of qantified brain iron by patient age were ths largely the inverse of the crves tracing the progressive decrease in in vivo T2 shortening with age. Becase T2 is so mch longer in aqeos soltion (5 milliseconds) than in the brain (7 milliseconds), we also stdied the effect of iron in agarose phantoms. These revealed mch lower T2 vales and a less dramatic decline in T2 with increasing iron concentration, bt one more comparable to the vales seen in patients. evertheless, both ferros and ferric irons led to a progressive decline in T2. Wismer et al (19) have demonstrated a similar effect of Fe34 on T2 in agar. Given the concrrent changes in iron and in T2 vales that occr with age and the similar effects of eqivalent concentrations of iron in vitro, it seems likely that the incremental accmlation of iron in certain gray matter strctres acconts for the progressive decrease in T2 relaxation time. We do not, however, sggest that all strctres with a low T2 nmber are rich in iron. Clearly, compact white matter pathways sch as the corps callosm, fornix, and anterior commissre have a short T2 signal that becomes progressively lower with the deposition and matration of the myelin (2, 21), bt not with the accmlation of iron. However, myelination occrs predominantly before the age of two and or data (Fig. 3) spport that T2 shortening in white matter does occr in a time frame similar to that for regions containing iron. After this age, however, T2 vales are relatively constant in white matter. Ths, it appears that the completion of myelination leads to constant T2 vales in white matter. In contrast, T2 vales contine to decrease in gray matter strctres, even after myelination has been completed, and in the same time frame as iron accmlation. Both or present qantitative stdy and or past investigations sing the Perl method for iron have shown a very low iron concentration in these heavily myelinated areas (21). This effect of mye- lin on T2 is independent of iron and may help to explain why Chen et al (6) failed to find a relationship between T2 shortening and iron concentrations when both white and gray matter strctres are considered together. When only gray matter strctres are analyzed, their data (6) do demonstrate a dramatic relationship between iron and T2 vales. This also has been conclded by another athor (4). It has been sggested that if only gray matter strctres are considered, the relationship between iron and T2 signal can be identified bt is less dramatic (4). Or reslts lead s to concr with the latter conclsion. We extend the work by Chen et al (6) by having done qantitative assessment in iron-containing anatomic regions beyond the basal ganglia and especially in ones that contain no significant amont of myelin (ie, S). In light of the low T2 nmbers of compact white matter fiber pathways, it is appropriate to ask whether myelin cold contribte to the low T2 signal in certain gray matter strctres. Myelinated bndles are prominent, for example, in the basal ganglia and R. It is of interest, therefore, that althogh or qantitative data show that both the S and R have similar amonts of iron, the latter strctre is darker on the T2- weighted image. Ths, myelin may contribte to the T2 shortening of certain gray matter strctres bt cannot accont for the progressive T2 shortening with age in the S, which contains little myelin. One limitation in or stdy is the difficlty in anatomically positioning the crsor in measring the T2 of the S and eliminating contamination of the nearby cerebral pedncle, which is myelinated and has T2 shortening. Or work leads s to conclde that there is an age-related accmlation of iron in five regions of the brain (R, S, D, CA, ptamen), correlating with an associated decrease in T2 vale from clinical scans, which can be also demonstrated in iron-containing phantoms. Brain iron appears to contribte to the progressive decrease of T2 signal that occrs with aging in these ironcontaining regions of the brain. References 1. Aoki S, Okada Y, ishimra K, et al. ormal deposition of brain iron in childhood and adolescence: MR imaging at 1.5T. Radiology 1989; 172: Drayer B, Brger P, Darwin R, Riederer S, Herfkens R, Johnson GA. Magnetic resonance imaging of brain iron. AJR: Am J eroradiol 1986;7: Drayer BP. Imaging of the aging brain. Part I. ormal findings. Radiology 1988; 166:

6 148 THOMAS AJR: 14, September/October Drayer BP. Basal ganglia: significance of signal hypointensity on T2- weighted MR images. Radiology 1989; 173: Brooks DJ, Lthert P, Gadian D, Marsden CD. Does signal-attenation on high-field T 2 -weighted MRI of the brain reflect regional cerebral Iron deposition? Observations on the relationship between regional cerebral water proton T 2 vales and iron levels. J erol erosrg Psychiatry 1989;52: I Chen JC, Hardy PA, Claberg M, et al. T2 vales in the hman brain: comparison with qantitative assays of iron and ferritin. Radiology 1989; 173: Fish WW. Rapid colorimetric micromethod for the qantitation of complexed iron in biological samples. Methods Enzymol 1988; 158: Gans A. Iron in the brain. Brain 1926;46: Harrison WW, etsky MG, Brown MD. Trace elements in hman brain: copper, zinc, iron, and magnesim. C/in Chim Acta 1968;21 : Schicha H, Kasperek K, Feinendegen LE, Siller V, Klein HJ. Eisen Konzentrationen in verschiedenen Abschnitten des menschlichen Gehirnes nd ihre Beziehngen zm Lebensalter. The iron content of hman brain and its correlation to age. Beitr Pathol Bd 1971;142: II. Ule G, Viilkl A, Berlet H. Sprenelemente im menschlichen Gehirn. II. Kper-, zink-, calcim- nd Magnesimkonzentration in 13 verschiedenen Hirnregionen wiihrend der 4. bis 8. Lebensdekade im vergleich zm Hirneisen. Z eroll974;26: Viilkl A, Berlet H, Ule G. Trace elements (C, Fe, Mg, Zn) of the brain dring childhood. eropadiatrie 1974;5: Hill JM, Switzer RC. The regional distribtion and celllar localization of iron in the rat brain. eroscience 1984; II : Hill JM. The distribtion of iron in the brain. In: Yodin MBH, ed. Brain iron: nerochemical and behavioral aspects. London: Taylor and Francis, 1988: Gomori JM, Grossman Rl. Mechanisms responsible for the MR appearance and evoltion of intracranial hemorrhage. Radiographies 1988;8: Grossman Rl, Gomori JM, Goldberg HI, et al. MR imaging of hemorrhagic conditions of the head and neck. Radiographies 1988;8: orfray JF, Coch JR, Elble RJ, Good DC, Manyam BV, Patrick JL. Visalization of brain iron by mid-field MR. AJR: Am J eroradiol 1988;9: Hallgren B, Sorander P. The effect of age on the non-haemin iron in the hman brain. J erochem 1958;3: Wismer GL, Bxton RB, Rosen BR, et al. Ssceptibility indced MR line broadening: applications to brain iron mapping. J Compt Assist Tomogr 1988;12: Barkovich AJ, Kjos BO, Jackson DE, orman D. ormal matration of the neonatal and infant brain: MR imaging at 1.5 T. Radiology 188;166: Crnes JT, Brger PC, Djang WT, Boyko OB. MR imaging of compact white matter pathways. AJR: Am J eroradiol1988;9:16168