proton magnetic resonance and human thyroid neoplasia i: discrimination between benign and malignant...

~ SPECIAL ARTICLE

Proton Magnetic Resonance and Human Thyroid Neoplasia I: Discrimination Between Benign and Malignant Neoplasms PETER RUSSELL, M.D., F.R.c.P.A., Camperdown, Australia, CYNTHIA L. LEAN, PH.D., LEIGH DELBRIDGE, M.D., F.R.A.C.S., St. Leonards andKogarah, Australia, GEORGE L. MAY, PH.D., SUSAN DOWD, CAROLYN E. MOUNTFORD, D.PHIL., Sydney, Australia

WRPOSE: Thyroid nodules are very common, yet the vast majority are biologically benign. The extreme difficulty facing the clinician selecting potentially malignant thyroid nodules for surgery was the subject of a recent editorial by Ernest L. Mazzaferri in the American Journal of Medicine (93:359-362, 1992). Here we evaluate the potential of proton magnetic resonance spectroscopy (lH MRS) to provide a solution to this problem.

PATIENTS: Thyroid tissue from fifty-three patients undergoing partial or total thyroidectomy for solitary thyroid nodules were assessed by lH MRS. RESULTS: When compared with the histologic diagnosis, lH MRS distinguished normal thyroid tissue (n = 8) from invasive papillary (n = 9), anaplastic (n = l), and medullary (n = 1) carcinomas with P values of c 0.0601, based on altered cellular chemistry. The same magnetic resonance (MR) criteria categorized path- ologically proven follicular carcinoma (n = 8) (established as such by the presence of capsular or vascular invasion at the periphery of the tumor, or by the presence of metastases in the patient) with the other thyroid cancers (P ~0.0001). All other “benign” follicular neoplasms (n = 34), including five atypical follicular adenomas, were assessed by the same lH MRS criteria and found to fit into one of the two above categories, viz. analogous to benign or malignant thyroid tissue.

CONCLUSIONS: Proton MRS has the potential to separate out a group of truly benign follicular neoplasms from follicular tumors (both follicular adenomas and follicular carcinomas) that have an atypical follicular pattern on cytologic examination. This is the first report of an objective diagnostic procedure that has the

From the Membrane MR Unit, Department of Cancer Medicine, University of Sydney, (SD, CLL, GLM, CEM), Department of Anatomical Patholonv. Roval Prince Alfred Hospital. Camoerdown. N.S.W.. (PR). Endocrine Surgical Units, Royal North Shore Hospital, St. Leonards] N.S.W.. and St. George Hospital, Kogarah, N.S.W., Australia.

This work was supported in part by a grant from the Leo and Jenny Leukaemia Cancer Foundation of Australia and the Australian National Health and Medical Research Council (Public Health).

Requests for reprints should be addressed to Carolyn E. Mountford, D.Phil., Membrane MR Unit, Department of Cancer Medicine, University of Sydney, N.S.W., 2006, Australia.

Manuscript submitted February 17, 1993, and accepted in revised form August 19, 1993.

potential to obviate surgical excision in a signficiant number of patients with benign follicular adenomas, independent of exhaustive histopathologic assessment.

T hyroid nodules are common and are estimated as being clinically evident in up to 10% of the popu-

lation.’ While the vast majority of these lesions are benign, either simple colloid nodules or benign follicular adenomas, the exclusion of thyroid ma&g- nancy remains a significant diagnostic problem, as recently described by Mazzaferri2 Patient age and sex, tumor size, and the presence of multiple nodules ah influence the relative risk of malignancy,3 but, currently, a diagnosis can only be made on biopsy ma- terial examined either cytologically or histologically. Even then, there are difficulties with making an accurate diagnosis. Fine needle aspiration biopsy (FNAEI) cytology, although accurate in differentiating papillary, medullary, and anaplastic carcinomas, is generally unable to differentiate genuinely benign from truly malignant follicular neoplasms.

Tumors that we arbitrarily designate as follicular adenomas and follicular carcinomas are indistinguishable in clinical, radiologic, and gross pathologic features. In the majority of cases, the cellular components of both types of tumor have the same histo- morphology. In such tumors, the criterion for mahg- nancy is the finding of capsular or vascular invasion at the periphery of the neoplasm. This requires surgical removal of the entire tumor, and extensive lab- oratory examination, which is rarely complete. For example, an average of eight paraffin blocks is rec- ommended to adequately examine the perimeter of a follicular tumor.4 As such, the diagnosis of malignancy depends very much on sampling, and is, at times, pure chance.

Even when the entire tumor capsule is examined, it is still not always possible to absolutely discriminate between benign and malignant follicular neoplasms. Different pathologists may classify the same lesion as either an atypical adenoma or an encapsu- lated follicular carcinoma, as seen in one patient in the series described herein. Thus, while the treatment of such tumors may be the same, the decision as to whether or not a patient has thyroid cancer may be- come a matter of the majority opinion.

The use of nuclear magnetic resonance imaging (MRI) to discriminate between normal and malignant

April 1994 The American Journal of Medicine Volume 96 383

PROTON MAGNETIC RESONANCE AND THYROID CARCINOMA/ RUSSELL ET /iL

tissue on the basis of relaxation rates of water sig- nals was proposed by Damadian5 20 years ago, but has proven to be a less-than-optimal approach.6 It is now understood that molecules other than water make a significant contribution to the MR profiles of tumors.7-g The concept that malignant cells produce a large number of lH MR visible molecules, few of which are observable in normal healthy tissues, has been explored as a diagnostic modality for cancer of the human uterine cervix,10-12 colon/rectum,13-16 breast,17 brain18 and prostate.1g?20 In the case of uterine cervix, the presence of a lipid spectrum, alone, is adequate to diagnose the presence of invasive cells in cervical punch biopsy with a P <O.OOOl. By contrast, for the human colon, which has a MR lipid signal from healthy mucosa, the overt increase in cellular metabolites facilitates the distinction between normal and carcinomatous tissue.15 In addition, by comparing colorectal cell lines of known de-differentiation it has been established that lH MRS can ob- jeetively grade human colorectal cancers ex vivo.16

in a 5-mm MRS tube so that the tissue would be positioned between the coils of the proton probe. Sufficient PBS/D,0 (400 pL) was used to cover the specimen. Having undergone lH MRS analysis, tissues were fixed in FAA (formalinacetic acidalcohol) and subsequently processed for routine histopathologic assessment.

This preliminary study was designed to investigate the potential for non-morphologic criteria (such as lH MRS), to categorize follicular thyroid neoplasms, and to help objectively confii a group of truly benign follicular neoplasms that are currently being submitted to surgical excision for diagnostic purposes.

Tissue was obtained at the time of surgery from 53 consecutive patients undergoing partial or total thyroidectomy for solitary thyroid nodules. The age range was 10 to 94 years (mean = 44.5) and the sex distribution 7~46 @IF). Indications for surgery in- cluded suspicion of malignancy on the basis of a malignant or atypical fine needle cytology report, or the presence of local symptoms. The fmal histologic di- agnoses in this group of patients were: benign follicular adenoma (20); Hurthle cell adenoma (5); hyper- plastic nodule (2); solitary colloid nodule (2); atypical follicular adenoma (5); follicular carcinoma (8); papillary carcinoma (9); anaplastic carcinoma (1); and medullary carcinoma (1). In addition, normal thyroid tissue was obtained from eight normal contralateral thyroid lobes as control specimens.

PATlENiS AND METHODS Patients

Methods

Preparation of tissue for MRS: Prior to the MRS experiment, specimens were washed with 5 x 1 mL PBS/D,O, and then placed on top of a glass wool plug

Magnetic resonance spectroscopy: All MR experiments were carried out on a Bruker wide-bore AM-360 spectrometer @-uker Analyusche Messtechnik [BAM], Karlsruche, Germany) (operating at 360 MHz) equipped with an Aspect 3000 computer @AM, Earls ruche, Germany). A standard 5-mm dedicated proton probehead @AM, Karlsruche, Germany) was used with the sample spinning at 20 Hz and the temperature maintained at 37°C. Residual water signal was sup- pressed by selective gated irradiation2i using low power (15 dI3). The chemical shifts of resonances were referenced to aqueous sodium 3-(trimethylsilyl)- propane-sulphonate (TSPS) at 0.00 ppm. One-dlmen- sional spectra were acquired over a sweep width of 3,957 Hz (10.0 ppm) using 8,192 data points, 128 ac- cumulations, an acquisition time of 1.14 seconds, and a relaxation delay of 2.00 seconds. T,- filtered 1D experiments were performed using a Carr-Purcell- Meiboom-Gill (CPMG) pulse sequence with an inter- pulse delay of z =l.OO msec and a delay between acquisitions of 1 second, as described in Mountford et al8 A line broadening of 3.0 Hz was applied to all 1D MR data prior to Fourier transformation. A ratio of the intensities of resonances at 0.9 ppm and 1.7 ppm were recorded. The symmetrized 2D correlated spectroscopy (COSY) (NS = 48, NE = 200) were processed using sine-bell and Lorentzian-Gaussian (LB = 30.0, GB = 0.20) window functions in the t, and G domains, respectively. The contour plot was gen- erated with the lowest level set close to the noise level and subsequent levels increasing by powers of two.

Tissue handling: Tissue specimens (3 mm3) obtained from excised lesions were placed in 1 mL PBS/D,0 in polypropylene vials, immediately im- mersed in liquid nitrogen, and subsequently stored at -70°C for less than 6 weeks.

Histo~athology: Tissue fixed in FXA, was paraf- fm embedded, sectioned at seven pm, and stained with hematoxylin and eosin according to standard protocols. In all follicular tumors not clinically malignant, but showing the features of atypical adenoma (nuclear atypia, numerous mitotic figures, disorga- nized architecture), a minimum of eight blocks were processed of tumor and capsule. The various pathologic subtypes of neoplasms were classified according to the current World Health Organization (WHO) classification of thyroid tumors22 and, where difficult diagnostic problems were encountered, as further described by Li Volsi.’

Assignment of lH MR Spectra One-dimensional (1D) Spectra: Prbton 1D and

T,- filtered (n2z =720 msec, where n = the number of

RESULTS

384 April 1994 The American Journal of Medicine Volume 96

3.0

I.0

2.0

3.0

4.0

5.0

6.0 I I I I I I 1 fvm

6.0 5.0 4.0 3.0 2.0 1.0 0.0

Figure 1. Proton magnetic resonance (MR) spectra (obtained at 360 MHz) of normal thyroid tissue: (A) Unfiltered onedimensional MR spectrum; (B) T,-filtered one-dimensional spectrum (CPMG, n27 =720 msec, NS =128) in which the intensity of resonances from molecules with short T,-relaxation values are reduced; (C) Symmetrized two-dimensional correlated spectroscopy (COSY) showing MR crosspeaks in two frequency dimensions, allowing identification of specific molecules. In the normal tissue, the spectra are dominated by lipid, with methyl (-CH,-1, methylene (-CH,-) and olefinic (-CH=CH-1 resonances at 0.9, 1.3, and 5.3 ppm, respectively. The diagnostic crosspeak at 1.7 ppm is denoted on one side of the diagonal.

n pulses and 2 =l msec) MR spectra of normal thyroid and papillary carcinoma are shown in Figures 1 and 2, respectively. The unfiltered 1D spectrum of normal thyroid (Figure 1A) is dominated by lipid with the methyl (-CHJ, methylene (-C$-) and olefinic (-CH=CH-) resonances at 0.9, 1.3, and 5.3 ppm, respectively. The spectrum from the papillary carcinoma does not have the characteristic lipid profie, but does contain many resonances including one at 3.2 ppm from the N-trimethyl (‘N(CH& of choline. Both spectra contain a prominent resonance at 1.7 ppm which is a composite from the -CH,- of lipid and -CH,- of lysine, as described below.

In the T,- filtered spectra, resonances from com- pounds with short T,- relaxation vahres such as lipid (-0.2 see) are reduced using a delay time (n2r) of 720 msec and those resonances from molecules with longer T,- relaxation values are more easily observed. Hence, in the spectra of the normal thyroid (Figure 1B) and the papillary carcinoma (Figure 2B) the lipid -CH,- (1.3 ppm) and -CH=CH- (5.3 ppm) resonances are significantly decreased, resulting in the exposure of N-trimethyl and other metabolite resonances assigned below.

Two-dimensional (2D) COSY Spectra Two-dimensional COSY reveals direct structural in-

formation by providing the coordinates of MR crosspeaks in two frequency dimensions. The off-diagonal crosspeaks show which resonances on the diagonal are scalar coupled and hence allow resonances to be assigned to specific molecules and quantification of the chemical changes that occur with disease sta- tus.23-25 The lH MR COSY spectra of normal thyroid and papillary carcinoma are shown in Figures 1C and 2C, respectively. Lipid resonances A to 6, which are dominant only in the spectrum of normal thyroid, were identified by May et al24 as triglyceride with crosspeaks A to F from the fatty acid chains and G and G’ from the vicinal and geminal protons on the glycerol backbonez5 (these connectivities are illus- trated elsewhere24). The methyl-methine couplings of alar-tine (1.49-3.79 ppm) and lactate anion (1.334.12 ppm), the methylene-methylene couplings of lysine (1.72-3.05 ppm) and choline (3504.07 ppm), and the methine-methine coupling of inositol(3.28--3.64 ppm) are also present in the spectra of both normal and carcinomatous tissues. Additional corsspeaks present in the spectrum of malignant tissue include those from free and bound ammo acids, glutamic acid/glu- tathione (2.21-2.62 ppm), histidine (3.22-3.95 ppm), leucine (0.97-1.78 ppm), threonine (1.334.27 ppm) and valine (1.03-2.34 ppm), and other species, including taurine (3.28-3.50 ppm) and bound fucose (1.334.27 ppm). Thus, from the COSY data, the prominent, diagnostically relevant resonance at 1.7 ppm in the 1D MR spectra can be seen to be comprised of a component from the -CH,- of triglyceride and lysine.

Diagnostic Resonances in IH MR Spectra Two major differences are evident when contrast-

ing 1D spectra from normal thyroid tissue and papillary carcinoma, viz the absence of lipid and the presence of the ammo acid metabolites in the carcinomatous tissue. The methyl resonances from ammo acids resonate at 0.90 ppm and the CH, from lipid resonates at 0.86 ppm. As a consequence, the intensity of the peak at 1.7 ppm to that at 0.9 ppm have


2.0

3.0

4.0

5.0

6.0 I I I I I I 1 fwm

6.0 5.0 4.0 3.0 2.0 1.0 0.0

Figure 2. Proton magnetic resonance spectra (obtained at 360 MHz) of papillary thyroid carcinoma: (A) Unfiltered one-dimensional spectrum; (B) T,-filtered one-dimensional spectrum (as described in Figure 1 legend); (C) Symmetrized two-dimensional spectrum. These spectra contain a prominent resonance at 1.7 ppm, which is a composite of the of -CH,- lipid and -CH,- of lysine. Both these crosspeaks are denoted on one side of the diagonal.

been measured in the unfiltered 1D MR spectrum and are presented in Figure 3 as a ratio. This ratio for normal thyroid tissue was greater than 1.1 for all specimens. In contrast, for specimens of papillary, medullary, and anaplastic carcinoma the ratio was consistently less than 1.1, with no overlap and a clear separation from the normal tissue (P <O.OOOl). This discrimination, then, between the MRS spectral char- acteristics of normal thyroid tissue and indisputably malignant thyroid tissue forms a basis upon which to attempt to discriminate follicular neoplasms.

AU cases of follicular thyroid neoplasms with un- equivocal histologic invasion of the capsule of adja- cent blood vessels and/or which were also manifestly clinically malignant, were categorized by lH MR along

PROTON MAGNETIC RESONANCE AND THYROID CARCINOMA/ RUSSELL ET AL

0.00 ’ NORMALS NEOPLASMS FOLLICULAR OTHER

(n=s) (ll=34) CARCINOMAS CARCINOMAS (tl=3) (n-11)

Figure 3. Plot of the ratio of the intensity of the resonance at 1.7 ppm (prominent composite resonance seen in the malignant biopsy spectra) and the resonance at 0.9 ppm (prominent methyl resonances observed in normal tissue) measured from unfiltered one-dimensional MR spectrum. Data are grouped on the basis of the final histopathology.

with the papillary, medullary, and anaplastic carcinomas, with a ratio less than 1.1 and were again separated from normal thyroid with a P <O.OOOl. Those follicular neoplasms diagnosed histologically as non- malignant were clearly separated into two groups, with one group having a spectral pattern with prop- erties directly comparable with normal thyroid tissue (Category A), and the other comparable with follicular, medullary, anaplastic and papillary carcinomas (Category B).

While there was no absolute distinction between the two groups on standard histologic criteria, a number of clinical correlates would appear to support the validity of this separation of “benign” follicular neoplasms into two groups. For example, there was a pre- dominance of very large solitary neoplasms (>50 mm) in Category B (4/20 versus O/14 in Category A) and a greater frequency of atypical follicular adenomas in Category B (4/20 versus l/14 in Category A). None of these clinical or pathologic criteria are, however, suf- ficiently discriminating to mandate a difference in patient management for follicular adenomas at this point in time, given our current understanding of neoplasia.

DISCUSSION We have demonstrated that lH MRS can separate

thyroid neoplasms of follicular type into two discrete categories. One category (benign/Category A) has an lH MR spectrum indistinguishable from that of normal thyroid tissue, while the other (malignant/Category B) is the spectral equivalent of papillary, anaplastic, and medullary carcinoma, as well as histologically proven follicular carcinomas. This categorization is based on changes to cell&r chemistry, specifically



increases in cellular metabolites and alterations to the lipid profile that occur in parallel with malignant transformation. One spectral change, used for diagnostic purposes, at 1.7 ppm, can be seen in the 2D COSY to be comprised of two major components viz the -CH,- from both the amino acid lysine (1.72405 ppm) and lipid (1.6-2.3 ppm). Both these crosspeaks are denoted in the 2D spectra While the lipid resonances (denoted A through G’) are clearly visible in the spectrum from normal thyroid (Figure lC), they are absent from the spectrum obtained from papillary carcinoma (Figure 2C). The -Cl!&- lysine crosspeak at (1.72405 ppm) is present in both spectra but is stronger, as denoted by the increased number of contour levels in the crosspeak, in the carcinoma spectrum. Thus, it would appear that, during the biologic transition from normal to carcinomatous, both the MR measurable lipid and lysine levels are altering. We must, therefore, conclude that both the lipid and the lysine have potential discriminatory power to aid in the correct diagnosis of follicular adenoma.

Fagin2’j recently proposed a pattern of sequential mutational events that may underlie initiation and progression of thyroid neoplasia. Whether the changes in cellular chemistry seen in this series of tumor represents progression towards malignant transformation, similar to that seen in colon tumorigenesis,27 will require further spectral analysis using more sophisticated mathematical techniques.

Two questions now arise: (1) Can follicular neoplasms that have a benign MRS spectrum be conti- dently considered as biologically benign, thus re- moving the need for surgical excision? (2) Can those follicular neoplasms, which have been diagnosed as benign on orthodox histologic grounds, but which demonstrate a malignant change in their cellular chemistry by ‘H MRS, actually have the potential to metastasize? Given the interventionist nature of the diagnostic approach to follicular thyroid neoplasms, it is unlikely that the latter question can be answered conclusively, even with a large cohort studied over many years.

With respect to the first question, however, we be- lieve that this study clearly identifies a group of follicular neoplasms that are benign. There has not been a single malignant tumor, either papillary, follicular, anaplastic or medullary, which has shown anything other than the “malignant/Category B” spectral pattern. The negative predictive value of this MRS test is 100%. However, until the MRS diagnosis can be con- firmed to be correct by independent means, the speci- ficity is only 52%.

The importances of this MRS analysis is the potential to obviate surgical excision in up to 33% of follicular neoplasms. Currently, all the tumors in this group will have been selected for surgery on the bases

of a FNAB showing a carcinoma or an “atypical” follicular pattern consistent with a follicular neoplasm. As cytologic criteria, alone, cannot presently distin- guish a follicular carcinoma from a benign follicular adenoma, all such patients undergo surgery for diagnostic purposes. Since, in 53 consecutive, solitary thyroid nodules in this study, all malignant tumors clearly demonstrated a “malignant/Category B” spectral pattern, and no patient with a “benign/Category A” pattern had any histopathologic evidence of malignancy whatsoever, we consider that there is sufficient data to plan a clinical trial of conservative management of patients with follicular neoplasms based on magnetic resonance spectroscopy. The problems of validating the MR method will be similar to those encountered with FNAB, when fmt introduced. Proof of both these methods await long-term clinical trial confirmation to verify the correct pathologic diagnosis and patient outcome.

The signiticance of the “malignant/Category B” spectral pattern in those examples of histologically benign follicular adenoma remains unclear. On the basis of substantial clinical evidence such neoplasms are likely to follow a benign clinical course, regard- less of the nature of their fundamental biology. On the one hand, late me&stases from “benign” atypical follicular adenomas are exceedingly rare, and such instances presumably relate to sampling errors in the original specimen. On the other hand, some follicular carcinomas, which show evidence only of capsular penetration but no vascular invasion, do not result in either distant metastases, or death, when followed for long periods of time, and are, thus, biologically very low-grade malignancies.28

Experiments are currently underway to ascertain if the same spectral patterns are technically obtain- able from fine-needle aspirates of thyroid lesions. If successful, this would enable the selection of patients for conservative management by FNAB, thus avoid- ing surgical excision in the first instance in those patients with a “benign/Category A” spectral pattern.

However, it must be said that, if MRS can accu- rately diagnose follicular neoplasia on the basis of FNAB, the most valuable use of this technology will be to replace cytopathology as the current primary screening test to obviate surgical excision simply for diagnostic purposes. Other future experiments include in vivo spectroscopy of the thyroid and a more rigorous mathematical analysis of the spectral data obtained thus far. It is hoped such an analysis will further refine the discriminatory capacity of ‘H MRS with regard to thyroid neoplasms.

ACKNOWLEDGMENT We thank Professors I.C.P. Smith and M.H.N. Tattersall for their continued support and fruitful discussions; Wanda Mackinnon for helpful comments during



the writing of this report; Gordon Wong for his early contributions to this work; Judy Hood for typing the manuscript; Rebecca Hancock for help with the preparation of diagrams; and Dr. Ian Kalnins and Professor S. Reeve for access to tissue specimens.

REFERENCES 1. Li Volsi VA. Follicular lesions of the thyroid. In: Surgical Pathology of the Thyroid. Philadelphia: WB Saunders, 1990:173-212. 2. Mazzaferri EL. Thyroid cancer in thyroid nodules: finding a needle in the haystack. Am J Med. 1992;93:359-362. 3. Belfiore A, LaRosa GL, LaPorta GA, et al. Cancer risk in patients with cold thyroid nodules: relevance of iodine intake, sex, age, and multinodularity. Am J Med. 1992;93:363-369. 4. Yamashina M. Follicular neoplasms of the thyroid. Am J Surg Pathol. 1992;16:392-400. 5. Damadian R. Tumor detection by nuclear magnetic resonance. Science. 1971;171:1151-1153, 6. Bottomley PA. Human in tivo NMR spectroscopy in diagnostic medicine: clinical tool or research probe? Radiology. 1989;170:1-15. 7. Mountford CE, Wright LE, Holmes KT, et al. High-resolution proton nuclear magnetic resonance analysis of metastatic cancer cells. Science. 1984;226:1415-1418. 8. Mountford CE, MacKinnon WB, Delikatny EJ, Russell P. High resolution proton MRS in cancer pathology. In: de Certaines JD, Bovee WMMJ, Podo F, eds. Textbook of Magnetic Resonance Spectroscopy in Biology and Medicine. New York: Pergamon Press, 1992:507-526. 9. Mountford CE, Lean CL, MacKinnon WB, Russell P. The use of proton MR in cancer pathology. In: Webb GA, ed. Annual Reports on NMR Spectroscopy, Vol. 27. 1993:173-216. 10. Mountford CE, Delikatny EJ, Dyne M, et al. Uterine cervical punch biopsy specimens can be analysed by IH MRS. Mag Res Med. 1990;13:324-331. 11. Delikatny EJ, Russell P, Dyne M, et al. MRS distinguishes between carcinoma in situ and invasive carcinoma of the cervix. Proc Sot Mag Res Med. 1992;1:55. 12. Delikatny EJ, Russell P, Hunter JC, et al. Proton magnetic resonance and human cervical neoplasia. Ex viva spectroscopy distinguishes invasive carcinoma of the cervix from carcinoma in sihc and other preinvasive lesions. Radiology. 1993;188:791-796. 13. Mountford CE, Saunders JK, May GL, et al. Classification of human tumors

by high resolution magnetic resonance spectroscopy. Lancet. 1986;1:651-653. 14. Smith ICP, Princz EJ, Saunders JKS. Magnetic resonance spectroscopy in cancer research. J Can Assoc Radio/. 1990;41:32-38. 15. Lean CL, Newland RC, Ende DA, et al. Assessment of human colorectal biopsies by IH MRS: Correlation with histopathology. Mag Res Med. 1993;30:525533. 16. Mackinnon W, Lean C, Hancock R, et al. Cell surface fucosylation reflects loss of cellular differentiation in human colorectal cancer. Proc Sot Mag Res Med. 1992;2:3531. 17. Roman S. A ‘H MRS study of breast cancer. BSc Med Thesis, University of Sydney, 1992. 18. Rutter A, Smith ICP, Saunders JK, Hugenholtz H. Characterization of human brain tumors by proton MRS. Proc Sot Mag Res Med. 1991;1:1065. 19. Moats R, Jones L, Gravely L, et al. Stability of 1-D and 2-D COSY NMR on fresh human prostate samples. Proc Sot Mag Res Med. 1992;2:3619. 20. Sanders JA, Thomasson DM, Sillerud LO. l-Dimensional proton chemical shift imaging of the human prostate in viva. Proc Sot Mag Res Med. 1992;2:3305. 21. Jesson JP, Meakin P, Kneissell G. Homonuclear decoupling and peak elimination in Fourier transform nuclear magnetic resonance. J Am Chem Sot. 1973;95:618-620. 22. Hedinger C, Williams ED, Sobin LH. WHO histological classification of thyroid tumors: a commentary on the 2nd edition. Cancer. 1989;63:908-911. 23. Cross K, Holmes KT, Mountford CE, Wright PE. Assignment of acyl chain resonances from membranes of mammalian cells by two-dimensional NMR methods. Biochemistry. 1984;23:5895-5897. 24. May GL, Wright LC, Holmes KT, et al. Assignment of methylene proton resonances in NMR spectra of embryonic and transformed cells to plasma membrane triglyceride. J Biol Chem. 1986;261:3048-3053. 25. Holmes KT, Mountford CE. Identification of triglyceride in malignant cell membranes. J Mag Res. 1991;97:407-409. 26. Fagin JA, Matsuo K, Karmaker A, et al. High prevalence of mutations of p53 gene in poorly differentiated human thyroid carcinomas. J Clin Invest. 1993;91:179-184. 27. Fearon ER, Vogelstein B. A genetic model for colorectal tumorigenesis. Cell. 1990;61:759-767. 28. van Heerden JA, Hay JE, Goellner JR, et al. Follicular thyroid carcinoma with capsular invasion alone: a non-threatening malignancy. Surgery. 1992;112:1130-1138.


proton magnetic resonance and human thyroid neoplasia i: discrimination between benign and malignant...

Documents