Detection of AA76, a Common Form of Amyloid A Protein, as a Way of Diagnosing AA Amyloidosis

Transcription

1 Available online at Annals of Clinical & Laboratory Science, vol. 46, no. 2, Detection of AA76, a Common Form of Amyloid A Protein, as a Way of Diagnosing AA Amyloidosis Junji Sato 1,2, Yasuaki Okuda 3, Takeshi Kuroda 4, and Toshiyuki Yamada 1 1 Department of Clinical Laboratory Medicine, Jichi Medical University, Tochigi, 2 Biochemical Research Laboratory, Eiken Chemical Co., Ltd., Tochigi, 3 Centre for Rheumatology, Dohgo Spa Hospital, Ehime, and 4 Health Service Center, Niigata University, Niigata, Japan Abstract. Reactive amyloid deposits consist of amyloid A (AA) proteins, the degradation products of serum amyloid A (SAA). Since the most common species of AA is the amino terminal portion produced by cleavage between residues 76 and 77 of SAA (AA76), the presence of AA76 in tissues could be a consequence of AA amyloid deposition. This study assessed the diagnostic significance of the detection of AA76 for AA amyloidosis using two different approaches. Biopsy specimens (n=130 from 54 subjects) from gastroduodenal mucosa or abdominal fat (n=9 from 9 subjects) of patients who had already been diagnosed with or were suspected of having AA amyloidosis were used. Fixed mucosal sections were subjected to immunohistochemistry using a newly developed antibody recognizing the carboxyl terminal end of AA76 (anti-aa76). The non-fixed materials from gastroduodenal mucosa or abdominal fat were subjected to immunoblotting for detection of the size of AA76. Among the gastroduodenal specimens (n=115) from already diagnosed patients, the positive rates of Congo red staining, immunohistochemistry using anti-aa76, and immunoblotting were 68.4%, 73.0%, and 92.2%, respectively. The anti-aa76 did not stain the supposed SAA in the blood or leakage, which was stained by anti-saa antibody. AA76 was not detected either by immunohistochemistry or by immunoblot in the materials from patients in whom AA amyloidosis had been ruled out. In the abdominal fat, the immunoblot detected AA76 in 8 materials from 8 already diagnosed patients and did not in 1 patient whose gastroduodenal mucosa was negative. In conclusion, the detection of AA76 may alter the ability to diagnose AA amyloidosis. In immunohistochemistry for fixed specimens, the new anti- AA76 antibody can improve the specificity. Immunoblot for non-fixed materials, which can considerably improve the sensitivity, should be beneficial for small materials like abdominal fat. Key words: AA amyloidosis, abdominal fat, immunoblot, monoclonal antibody, serum amyloid A. Introduction Reactive amyloidosis is a serious complication of chronic inflammatory diseases such as rheumatoid arthritis and tuberculosis, in which amyloid A (AA), the product of the degradation of serum amyloid A (SAA) proteins, forming insoluble fibrillar aggregates, is deposited in the extracellular spaces of miscellaneous organs [1,2]. This disease is diagnosed histologically using biopsy specimens taken mainly from gastrointestinal mucosa [3]. Positive Congo red staining together with clinical information is usually effective for its diagnosis, and immunohistochemistry using antibodies to SAA strongly supports the diagnosis. Address correspondence to Toshiyuki Yamada, MD, Department of Clinical Laboratory Medicine, Jichi Medical University, 3311 Yakushiji, Shimotsuke, Tochigi , Japan; phone: +81 (0) ; fax: +81 (0) ; e mail: toshiyam@jichi.ac.jp The disease has long been thought to be progressive, but recent advances in anti-inflammatory therapies including anti-cytokine regimens have successfully reduced inflammatory activity and often resulted in remission of the disease via reduction of the amyloidotic precursor SAA [4,5]. Consequently, the detection of amyloid deposits in a smaller mass has been required. Congo red staining, the standard method for detecting amyloid deposits, sometimes misses very small deposits. Immunohistochemistry is sufficiently sensitive to detect such deposits, but a positive reaction does not always indicate amyloid deposits since the antibodies used may react with precursor SAA. Thus, methods that can specifically detect amyloidotic AA with high sensitivity and specificity are required. AA proteins are the amino terminal segments of SAA, which are the result of limited proteolysis /16/ by the Association of Clinical Scientists, Inc.

2 148 Annals of Clinical & Laboratory Science, vol. 46, no. 2, 2016 Table 1. Diagnostic performance of Congo red staining, immunohistochemistry, and the detection of AA76 band by immunoblot. Positive number (%) is shown. Positive No. (%) Group No. IHC with IHC with Congo red anti-saa30 CF6 Immunoblot Negative* 15 0 (0%) 0 (0%) 0 (0%) 0 (0%) Second look Specimens (68.4%) 84 (73.0%) 84 (73.0%) 106 (92.2%) Subjects* (84.6%) 36 (92.3%) 36 (92.3%) 39 (100%) *Positivity was defined when one or more specimen was positive. IHC: immunohistochemistry. Table 2. Diagnostic performance of Congo red staining and the detection of AA76 band by immunoblot in abdominal fat specimens. Case Congo red Immunoblot Grade (0-3) of amyloid deposits in gastroduodenal mucosa 1 IM IM NA NA - 0 IM: insufficient material, NA: not available. Because of multiple cleavage sites, AAs vary in length [6-10]. The most common site of cleavage is between the 76 th and 77 th residues of SAA. The resulting amino terminal part, named AA76 here, is the most abundant among AAs in the deposits [11]. However, this has not been confirmed precisely, although AA76 is hardly detected in plasma or tissues in non-amyloidotic subjects. If such confirmation were achieved, the detection of AA76 may be directly linked to a specific diagnosis of reactive amyloidosis. In this study, we adapted two different approaches for detecting AA76. One is immunohistochemistry using a newly established antibody specific to AA76. The other is immunoblotting, which detects the AA76 size in non-fixed specimens. Materials and Methods Patient samples. The study protocol was approved by the ethics committee of Jichi Medical University, Dohgo Spa Hospital, and Niigata University. Biopsy samples obtained endoscopically from gastroduodenal mucosa of rheumatoid arthritis (RA) patients in Dohgo Spa Hospital or Niigata University Hospital after obtaining informed consent were used in this study. The study included 54 females, aged 58 to 84 years (mean: 68.1). Fifteen patients received a biopsy for the first time to screen for the presence of amyloid deposits. From this screening, all of the results were negative, and these subjects are referred to as the negative group. The other 39 had already been diagnosed with AA amyloidosis and underwent a biopsy to evaluate the amyloid load. These are referred to as the second look group. Fat tissues were aspirated from the abdominal subcutaneous spaces as previously described [12]. Fat samples were taken from 1 patient among the negative group and 8 patients among the second look group as mentioned above. Antibodies. A previously established monoclonal antibody, anti-saa30, which was confirmed to recognize a portion around residue 30 of SAA [13], was used. In this study, antibodies recognizing the carboxyl terminus of AA76 were newly established. A peptide having Ser76 (paa76: DGDGDHGAEDS) or Ala78 (paa78: DGDGDHGAEDSLA) at the carboxyl terminus was synthesized with the incorporation of soluble spacer. For immunization, KLH, at the same weight as paa76, was conjugated at the amino terminus of the peptide, while paa76 and paa78 were conjugated with bovine serum albumin (BSA) and used for the examination of antibody reactivity. Fifty micrograms of the peptide, together with an equal volume of Freund s complete adjuvant, was immunized in anesthetized WKY rat (Charles River Laboratories, Japan). Two weeks later, the iliac lymph nodes were removed [14]. Collected lymph node lymphocytes and murine myeloma cells were assessed for the

3 Detection of AA76 for AA amyloidosis 149 Figure 1. Sensorgram of the interaction of monoclonal antibodies with paa76-bsa and rsaa1.1 on BIAcore. An overlay of sensorgrams obtained with anti-saa30 (dashed line) and CF6 (solid line) is shown. Monoclonal antibodies were immobilized onto a CM5 sensor chip, and paa76-bsa diluted with running buffer was injected for2min (left). Similarly, rsaa1.1 was injected (right). The response from the blank flow cell was subtracted from that of the antibody surface. Data are expressed as resonance units (RU). apparatus was operated according to the manufacturer s standard protocols. The antibodies, at a concentration of approximately 50 μg/ml, were immobilized covalently onto CM5 sensor chips via amine coupling as the amount of immobilization reached 10,000 RU. As an analyte, paa76-bsa or recombinant SAA1.1 [17] dissolved, at 1 μg/ml or 5 μg/ml, respectively, in a running buffer, HBS-EP (10 mm HEPES ph7.4, 150 mm NaCl, 3 mm EDTA, and 0.005% polysorbate 20), was applied to the chip at a flow rate of 20 μl/min. The response from a blank flow cell was subtracted from that of the antibody surface. Figure 2. Immunoblot for evaluation of the specificity of the newly established anti-aa76 antibody CF6. Samples were 0.1 μg of recombinant AA76 (lanes 1 and 3) and 0.1 μg of recombinant SAA1.1 (lanes 2 and 4). Anti-SAA antibody was reacted with lanes 1and 2, while CF6 was reacted with lanes 3and 4. CF6 did not recognize SAA. Molecular weight (Daltons) is indicated on the left side. establishment of a hybridoma producing the desired monoclonal antibodies, according to a previously described method [15,16]. Hybridoma was selected by the reactivity of culture supernatant in enzyme-linked immunosorbent assay (ELISA). A positive reaction was found with the immunized peptide paa76 and a negative reaction with the longer peptide paa78. Established clones were grown in large-scale culture, and immunoglobulin fractions were obtained from the supernatant by anion exchange chromatography using QAE resin (Tosoh, Japan) and used as monoclonal antibodies. Characterization of monoclonal antibodies by SPR analysis. The specificity of newly developed antibodies was assessed by surface plasmon resonance analysis using a BIAcore 2000 apparatus (GE Healthcare, UK). The Sample processing. In the endoscopy procedures, two materials were obtained from each of the antrum of the stomach, the bulb portion of the duodenum, and the second portion of the duodenum. One was treated with 10% formalin and processed to paraffin-embedded sections. The other was kept frozen at -70 º C until immunoblot analysis. Half of the abdominal fat was attached on a glass slide, dried, and treated with 10% formalin followed by Congo red staining. The other half was frozen at -70 º C until immunoblot analysis. Histological analysis. Gastroduodenal specimens and abdominal fat specimens were subjected to conventional Congo red staining. Immunohistochemistry was performed for gastroduodenal specimens. Briefly, they were treated with 10% formic acid for 10 minutes and subjected to the avidin biotin peroxidase complex method using a Vectastain ABC kit (Vector Laboratories, USA). As primary antibodies, anti-saa30 or a newly established one was incubated at 2 μg/ml with the specimens for 60 minutes at room temperature, followed by the procedures according to the manufacturer s protocol.

4 150 Annals of Clinical & Laboratory Science, vol. 46, no. 2, 2016 Figure 3. Immunohistochemistry for gastroduodenal mucosa of a patient from the second look group. Anti-SAA30 (left) or CF6 (right) was used. Original magnification was x40. The amount of the amyloid was graded 0~3 as follows. Grade 0: negative, Grade 1: sparse and tiny, Grade 2: between Grade 1 and Grade 3, Grade 3: massive (>30% of the specimen). Immunoblot. Non-fixed tissues (approximately 1 mg at most), either gastroduodenal mucosa or abdominal fat, were homogenized in 100 μl of 6 M urea and 0.1 M Tris-HCl, ph 8.5. After being combined at a ratio of 1:1 with a sodium dodecyl sulfate (SDS) electrophoresis buffer containing dithiothreitol (DTT) and heated, 10 μl of each sample was subjected to tricine SDS polyacrylamide gel electrophoresis, followed by being electrotransferred to a polyvinyl difluoride membrane (PVDF). The membrane was blocked with 1% bovine serum albumin (BSA) in phosphate-buffered saline (PBS), ph 7.5, and was then incubated with anti-saa30 at 1 μg/ml in 1% BSA-PBS containing 0.05% Tween 20 (BSA-PBS-T) for one hour at room temperature. After washing, peroxidase-conjugated anti-rat immunoglobulin (Tago, USA) at 1:1000 dilution in BSA-PBS-T was reacted for one hour at RT. After a final wash, the chemiluminescence was developed using the substrate EzWestLumi plus (ATTO, USA) and read using an apparatus, Ez-Capture MG (ATTO). The AA76 band was defined by agreement with the electrophoretic mobility of recombinant AA76, which had been produced in E. coli [18]. Results Evaluation of monoclonal antibodies. The clone CF6 was finally established. In SPR analyses, CF6 reacted well with paa76-bsa but not with rsaa1.1, while the reverse was shown for anti- SAA30 (Figure 1). In immunoblot analyses, CF6 reacted with AA76 but not with rsaa1.1. Its reactivity was considerably weaker than that of anti- SAA30 (Figure 2). Immunohistochemistry for gastroduodenal specimens with the new antibody. Both anti-saa30 and CF6 positively stained all of the Congo red-positive amyloid deposits in the gastroduodenal specimens (Figure 3). Notably, the anti-saa30 stained nonamyloid substances, probably because the precursor SAA leaked from the circulation, while CF6 did not (Figure 4). On the other hand, in the negative group, both anti-saa30 and CF6 yielded negative results (Table 1). Immunoblot. In immunoblot for all of the materials that were positive by immunohistochemistry, anti- SAA30 reacted with a band identical in size to raa76 and the other high-molecular-weight band, probably a dimeric form of AA76 (Figure 5, lane 4). In addition, for the materials for which the corresponding specimens were negative by immunohistochemistry, immunoblot showed a weakly positive result (Figure 5, lane 3). Some materials that showed positive staining in non-amyloid substances, as shown in Figure 4, had a positive band corresponding to intact SAA (Figure 5, lane 5). On the other hand, in the negative group, the band corresponding to AA76 was not detected (Figure 5, lane 2). Diagnostic performance. In 115 gastroduodenal mucosa samples from the second look group (n=39), the positive numbers (%) for Congo red, immunohistochemistry, and immunoblot were 78(68.4%), 84(73.0%), and 106(92.2%), respectively. Individual subjects were defined as positive when having at least one positive sample, the positive numbers (%) for Congo red, immunohistochemistry, and immunoblot were 33(84.6%), 36(92.3%), and 39(100%), respectively (Table 1)..

5 Detection of AA76 for AA amyloidosis 151 Figure 4. Immunohistochemistry for gastroduodenal mucosa of a patient from the second look group. The amyloid deposits might have disappeared in this patient. Anti-SAA30 (left) stained non-amyloidotic materials while CF6 (right) did not. Original magnification was x40. In abdominal fat, immunoblot was positive in all 8 materials from the amyloidosis-positive patients and was negative in a sample from a subject in whom amyloidosis had been ruled out (Table 2). Of note, immunoblot detected the AA76 band in the materials for which the specimens for Congo red staining could not be generated because of insufficient sampling. Discussion AA amyloidosis involves chronic inflammatory diseases, mainly RA in Japan. It is believed that the poor control of RA results in prolonged hyperexpression of SAA and leads to AA amyloidogenesis. A precursor-product relationship between SAA and deposited AA is generally accepted, but the significance of limited degradation of SAA is still under debate. In a mouse model of this disease, the conversion of SAA to AA is believed to be a post-deposition phenomenon [19]. However, in human diseases, precursor SAA has hardly been detected in the deposits. In addition, the most common species of AA proteins is AA76, which results from proteolysis between residues 76 and 77 in an SAA molecule. Therefore, the conversion of SAA to AA76 may be required for fibril formation if it occurs before or during fibril formation. Otherwise, it may result from incomplete absorption by the host if it occurs after fibril deposition. Irrespective of which of these mechanisms is involved, the detection of AA76 would be expected to be linked directly to the presence of amyloid deposits. Although AA species other than AA76 have been detected in AA fibrils [9,10], we believe that AA deposits without AA76 are ultimately unlikely. In addition, to satisfy the diagnostic specificity, AA76 has to appear only in the process of amyloid deposition. In this study, AA76 was not detected in the materials from patients without amyloidosis. It has not been detected in plasma in our experience. It is likely that AA76 is highly specific to amyloidogenesis; however, further cases should be accumulated. Generally, the diagnosis of AA amyloidosis is not difficult. Congo red staining is usually effective for this purpose. When Congo red has failed to detect the deposits for some reason, immunohistochemistry is always supportive, as presented in this paper. The new antibody CF6 showed complete identity with anti-saa30 in the staining of amyloid deposits. This suggests that AA76 is always present in the deposits. CF6 can provide more specific staining for amyloid deposits than anti-saa30 since it does not react with non-amyloidotic SAA. The amyloid deposits of a patient, from whom Figure 4 specimen was taken, seemed to disappear as judged from the immunoblot result (Figure 5, lane 5). Although skilled pathologists might hardly misdiagnose the positive staining by anti-saa30 in this specimen as the amyloid, they might wish to avoid a noise. The use of CF6 in immunohistochemistry for diagnosing or following AA amyloidosis should seek a merit. Although a histological approach is generally useful, it has an intrinsic limitation. As shown in this study and a previous one [20], amyloid deposits are

6 152 Annals of Clinical & Laboratory Science, vol. 46, no. 2, 2016 of immunoblot indicate the retention of amyloid. Therefore, AA76 detection by immunoblot may overcome the limitation of histological approaches that the amyloid deposits happen to be missed in processed specimens. Figure 5. Immunoblot of gastroduodenal mucosa samples (lanes2to 5) from different patients. Anti-SAA30 reacted with the membrane. Lane 2: apatient from the negative group. Lane 3:apatient from the second look group whose immunohistochemistry was negative. Lane 4: apatient from the second look group whose immunohistochemistry is shown in Figure 3. Lane 5: a patient from the second look group whose immunohistochemistry is shown in Figure 4. Lane 1: recombinant AA76. Molecular weight (Daltons) is indicated on the left side. not often included in the processed specimen. When we obtained the specific antibody to AA76, our goal was to detect the amyloid as AA76 immunochemically using the new antibody in non-fixed materials. However, the new antibody, despite having high specificity to AA76, cannot be utilized because of its poor reactivity with degenerated AA76, as shown in the immunoblot results (Figure 1). In fact, the reactivity of CF6 markedly decreased when AA fibrils were treated with guanidine or urea (data not shown). The antibody may be directed to the epitope in the peptide used for immunization rather than that in the native form. Consequently, we abandoned the detection of AA76 in non-fixed materials by using the new antibody and adapted the immunoblot as it showed a band with a molecular size identical to that of AA76. The SDS-PAGE utilized in this study had separation performance sufficient to estimate the molecular size close to AA76 and recombinant AA76 was utilized for the identification of AA76, indicating the appropriateness of this strategy. Immunoblot was so sensitive that it detected AA76 in some specimens for which immunohistochemistry gave a negative result. On the basis of the history that the amyloid was positive once and there were positive results in at least one among three materials, it is appropriate that the positive results We now know that the immunochemical detection of AA76, such as by immunoblot, in nonfixed materials may be useful for detecting AA amyloid deposits. However, we may choose histological approaches for gastroduodenal mucosa for historical and practical reasons. This strategy should be adopted for non-fixed materials like abdominal fat. In fact, the AA amyloid load has been well evaluated in abdominal fat [21]. Our results were also effective to estimate the presence of deposits. Notably, AA76 was detected in the material for which a specimen for Congo red staining could not be made (Table 2). Even if a specimen for Congo red is available, evaluation of the staining is not always easy compared with that of ordinary tissue sections. In addition, the aliquot of the materials can be stored for further analysis, including proteomics technology [22]. Therefore, we propose that non-fixed materials like abdominal fat should not be fixed, but rather kept frozen until used in various applications. To establish the presented strategy definitively, diagnostic specificity should be further examined. Although AA76 was not detected in the patients without AA amyloidosis in this study, this should be confirmed by accumulating more cases. In conclusion, the detection of AA76 in biopsy materials can be another tool for diagnosing AA amyloidosis. In particular, its detection in nonfixed materials by immunoblot, which does not require a large volume of sample, should be considered as a method for screening AA amyloidosis because this method may reduce the burden of patients. Acknowledgment This work was supported by a Grant-in-aid for Scientific Research from the Ministry of Education, Science and Culture, Japan (No ), a grant to the Amyloidosis Research Committee for Research on Intractable Diseases from the Ministry of Health, Labor and Welfare, Japan, and a subsidy from JKA through its promotion funds from KEIRIN RACE.

7 Detection of AA76 for AA amyloidosis 153 References 1. Husby G, Marhaug G, Dowton B, Sletten K, Sipe JD. Serum amyloid A(SAA): biochemistry, genetics and the pathogenesis of AA amyloidosis. Amyloid Int J Exp Clin Invest 1994;1: Westermark GT, Fändrich M, Westermark P. AA amyloidosis:pathogenesis and targeted therapy. Annu RevPathol 2015;10: Kobayashi H, Tada S, Fuchigami T, Okuda Y, Takasugi K, Matsumoto T, Iida M, Aoyagi K, Iwashita A, Daimaru Y, Fujishima M. Secondary amyloidosis in patients with rheumatoid arthritis: diagnostic and prognostic value of gastroduodenal biopsy.br JRheumatol 1996; 35: Gillmore JD, Lovat LB, Persey MR, Pepys MB, Hawkins PN. Amyloid load and clinical outcome in AA amyloidosis in relation to circulating concentration of serum amyloid A protein. Lancet 2001; 358(9275): Okuda Y, Takasugi K. Successful use ofahumanized anti-interleukin-6 receptor antibody, tocilizumab, to treat amyloid A amyloidosis complicating juvenile idiopathic arthritis. Arthritis Rheum 20016;54: Levin M, Franklin EC, Frangione B, Pras M. The amino acid sequence of a major nonimmunoglobulin component of some amyloid fibrils. JClin Invest 1972;51: Sletten K, Husby G. The comolete amino acid sequence of nonimmunoglobulin amyloid fibril protein AS in rheumatoid arthritis. Eur JBiochem 1974;41: Husebekk A, Skogen B, Husby G,Marhaug G. Transformation of amyloid precursor SAA to protein AA and incorporation in amyloid fibrils in vivo. Scand JImmunol 1985;21: Moyner K, Sletten K, Husby G, Natvig JB. An unusually large (83 amino acid residures) amyloid fibril protein AA from a patient with Waldenstrom s macroglobulinemia and amyloidosis. Scand JImmunol 1980;11: Westermark P. The heterogeneity of protein AA in secondary (reactive) systemic amyloidosis. Biochim Biophys Acta 1982;701: Yamada T, Okuda Y. AA amyloid quantification in biopsy samples from stomach. Ann Clin Lab Sci 2012;42: Klemi PJ, Sorsa S, Happonen RP. Fine-needle aspiration biopsy from subcutaneous fat. An easy way to diagnose secondary amyloidosis. Scand JRheumatol 1987;16: Yamada T, Hirano N, Kuroda T, Okuda Y, Itoh Y: Generation and characterization of rat monoclonal antibodies against human serum amyloid A. Scand JImmunol 1997;46: Kishiro Y, Kagawa M, Naito I, Sado Y. A novel method of preparing rat-monoclonal antibody-producing hybridomas by using rat medial iliac lymph node cells. Cell Struct Funct 1995;20: Galfre G, Milstein C. Preparation of monoclonal antibodies: strategies and procedures. Methods Enzymol 1981; 72: Schmitt JJ, Zimmermann U, Neil GA. Efficient generation of stable antibody forming hybridoma cells by electrofusion. Hybridoma 1989;8: Yamada T, Kluve-Beckerman B, Liepnieks JJ, Benson MD: Fibril formation from recombinant human serum amyloid A. Biochim Biophys Acta 1994;1226: Yamada T, Liepnieks JJ, Kluve-Beckerman B, Benson MD. Cathepsin Bgenerates the most common form of amyloid A (76 residues) as a degradation product from serum amyloid A. Scand JImmunol 1995; 41: Shiroo M, KawaharaE, Nakanishi I, Migita S. Specific deposition of serum amyloid A protein 2in the mouse. Scand J Immunol 1987; 26: Okuda Y, Takasugi K, Oyama T. Amyloidosis in rheumatoid arthritis-clinical study of 124 histologically proven cases. Ryumachi 1994; 34: Hazenberg BP, Bijzet J, Limburg PC, Skinner M, Hawkins PN, Butrimiene I, Livneh A, Lesnyak O, Nasonov EL, Filipowicz- Sosnowska A, Gül A, Merlini G, Wiland P, Ozdogan H, Gorevic PD, Maïz HB, Benson MD, Direskeneli H, Kaarela K, Garceau D, Hauck W, Van Rijswijk MH. Diagnostic performance of amyloid A protein quantification in fat tissue of patients with clinical AA amyloidosis. Amyloid J Prot Fold Dis 2007; 14: Vrana JA, Thesis JD, Dasari S, Mereuta OM, Dispenzieri A, Zeldenrust SR, Gertz MA, Kurtin PJ, Grogg KL, Dogan A. Clinical diagnosis and typing of systemic amyloidosis in subcutaneous fat aspirates by mass spectrometry-based proteomics. Haematologica 2014; 99: