hATTR Pathology: Nerve Biopsy Results from Italian Referral Centers

Abstract: Objective Pathological evidence of amyloid on nerve biopsy has been the gold standard for diagnosis in hereditary transthyretin amyloidosis polyneuropathy (hATTR-PN) for a long time. In this article, we reviewed the pathological findings of a large series of sural nerve biopsies from a cohort of hATTR-PN patients, collected by different Italian referral centers. Patients and Methods: We reviewed clinical and pathological data from hATTR-PN patients, diagnosed and followed in five Italian referral centers for peripheral neuropathies. Diagnosis was formulated after a positive genetic test for transthyretin (TTR) mutations. Sural nerve biopsy was performed according to standard protocols. Results: Sixty-nine sural nerve biopsies from hATTR-PN patients were examined. Congo red positive deposits were found in 73% of cases. Only the Phe64Leu mutation failed to show amyloid deposits in a high percentage of biopsies (54%), as already described. Unusual pathological findings, such as myelin abnormalities or inflammatory infiltrates, were detected in occasional cases. Conclusions: Even if no longer indicated to confirm hATTR-PN clinical suspicion, nerve biopsy remains, in expert hands, a rapid and inexpensive tool to detect amyloid deposition. In Italy, clinicians should be aware that a negative biopsy does not exclude hATTR-PN, particularly for Phe64Leu, one of the most frequent mutations in this country.


Statistical Analysis
Statistical analysis of the data was performed by SPSS (Statistical Package for Social Science) version 24.0. Fisher's 2-tailed exact test was used to compare numerical and nominal dichotomous variables, respectively. Significance was set at 0.05.

Ethics
Nerve biopsies are diagnostic tools in the setting of neuropathies, and all patients signed a written informed consent before the procedure. The study conforms to the ethical guidelines of the 1975 Declaration of Helsinki (6th revision, 2008), as reflected in a priori approval by the institution's human research committee.

Clinical Results
A total of 67 patients (54 males and 13 females) underwent a sural nerve biopsy. The male-to-female ratio was 4.2. The mean age at disease onset was 64.1 years (median 67.0, standard deviation 10.8, range . Seven patients showed an early onset of the disease (<50 years). Mean disease duration at the time of biopsy was 2.9 years (median 2.0, standard deviation 3.1).
The geographical distribution of patients covered the whole Italy; 22 patients were from Sicily, 12 patients were from Lazio, 6 patients were from Apulia, 5 patients were from Calabria, 4 patients were from Campania and Lombardy, 3 patients were from Piedmont and Abruzzo, 2 patients were from Veneto and 1 patient was from Alto-Adige, Liguria, Emilia-Romagna and Tuscany. Two patients came from abroad (Macedonia and Holland, respectively).
Clinical disabilities, assessed by PND score, were as follows: 1 patient was in stage 1; 35 patients were in stage 2; 31 patients were in stage 3 (22 in stage 3A and 9 in stage 3B).
Considering NCS, an axonal neuropathy was present in 54 patients, a mixed neuropathy in 11 patients and a demyelinating neuropathy in 2 patients. NCS were performed 1-6 months before the sural nerve biopsy.

Pathological Results
A total of 69 sural nerve biopsies were analyzed. Two male patients (Phe64Leu and Glu62Lys) underwent a second nerve biopsy, two years after the first one in both cases. The mean age at biopsy was 67.0 years (median 69.0, standard deviation 10.0, range 40-83).
All biopsies showed evidence of axonal loss, which was severe in 42 cases, moderate in 15 and slight in 4 (Figure 1a,b). In 8 biopsies, we observed a complete loss of myelinated fibers ( Figure 1b). Interestingly, in one biopsy, fiber loss was not homogeneously distributed among fascicles ( Figure 2a). In one further case, an intrafascicular inhomogeneous fiber loss was detected ( Figure 2b). Active axonal degeneration was found in 49 biopsies, with regenerating clusters in 22 (Figure 3a). Out of 22 biopsies, 5 showed regenerative clusters as possible evidence of a potential secondary cause of peripheral neuropathy.     Fibers surrounded by thin myelin sheaths were detected on semithin sections in 9 biopsies ( Figure 3a) and were confirmed on teased fiber examination, which was available only for 36 biopsies, in 3 cases (Figure 3b,c). One further case showed only segmental demyelination on teased fiber analysis. Inflammatory infiltrates, composed of histiocytes and some T lymphocytes, were observed in four biopsies (Figure 4a,b). In one of these biopsies (the patient with the Val32Ala mutation), a further possible secondary cause of polyneuropathy, namely renal failure, was also present.   Fibers surrounded by thin myelin sheaths were detected on semithin sections in 9 biopsies ( Figure 3a) and were confirmed on teased fiber examination, which was available only for 36 biopsies, in 3 cases (Figure 3b,c). One further case showed only segmental demyelination on teased fiber analysis. Inflammatory infiltrates, composed of histiocytes and some T lymphocytes, were observed in four biopsies (Figure 4a,b). In one of these biopsies (the patient with the Val32Ala mutation), a further possible secondary cause of polyneuropathy, namely renal failure, was also present. Fibers surrounded by thin myelin sheaths were detected on semithin sections in 9 biopsies ( Figure 3a) and were confirmed on teased fiber examination, which was available only for 36 biopsies, in 3 cases (Figure 3b,c). One further case showed only segmental demyelination on teased fiber analysis.
Inflammatory infiltrates, composed of histiocytes and some T lymphocytes, were observed in four biopsies (Figure 4a,b). In one of these biopsies (the patient with the Val32Ala mutation), a further possible secondary cause of polyneuropathy, namely renal failure, was also present.  Congo red staining was positive in 50 of 69 cases (72.5%) (Figure 5a,b). Comparing the two most common mutations in our cohort, Congo red positive deposits were found in 21 of 24 Val30Met cases versus 13 of 27 Phe64Leu cases (p = 0.0035). We did not find any significant association between Congo red positive deposits and the clinical severity of polyneuropathy assessed by PND score, neither in the entire cohort (PND 1 and 2: 26 Congo red positive and 11 Congo red negative vs. PND 3A and 3B: 24 Congo red positive and 8 Congo red negative), nor considering the two most Congo red staining was positive in 50 of 69 cases (72.5%) (Figure 5a,b). Comparing the two most common mutations in our cohort, Congo red positive deposits were found in 21 of 24 Val30Met cases versus 13 of 27 Phe64Leu cases (p = 0.0035). We did not find any significant association between Congo red positive deposits and the clinical severity of polyneuropathy assessed by PND score, neither in the entire cohort (PND 1 and 2: 26 Congo red positive and 11 Congo red negative vs. PND 3A and 3B: 24 Congo red positive and 8 Congo red negative), nor considering the two most frequent mutations (Val30Met PND 2: 9 Congo red positive and 3 Congo red negative vs. Val30Met PND 3A and 3B: 12 Congo red positive and 0 Congo red negative; Phe64Leu PND 2: 6 Congo red positive and 6 Congo red negative vs. Phe64Leu PND 3A and 3B: 7 Congo red positive and 8 Congo red negative). Congo red staining was positive in 50 of 69 cases (72.5%) (Figure 5a,b). Comparing the two most common mutations in our cohort, Congo red positive deposits were found in 21 of 24 Val30Met cases versus 13 of 27 Phe64Leu cases (p = 0.0035). We did not find any significant association between Congo red positive deposits and the clinical severity of polyneuropathy assessed by PND score, neither in the entire cohort (PND 1 and 2: 26 Congo red positive and 11 Congo red negative vs. PND 3A and 3B: 24 Congo red positive and 8 Congo red negative), nor considering the two most frequent mutations (Val30Met PND 2: 9 Congo red positive and 3 Congo red negative vs. Val30Met PND 3A and 3B: 12 Congo red positive and 0 Congo red negative; Phe64Leu PND 2: 6 Congo red positive and 6 Congo red negative vs. Phe64Leu PND 3A and 3B: 7 Congo red positive and 8 Congo red negative). Immunohistochemistry with anti-TTR, anti-kappa light chains and anti-lambda light chains was performed in 35 out of 50 Congo red positive biopsies, and a binding with anti-TTR antibodies was observed in all cases (Figure 6a-c). Interestingly, four cases also showed a cross-reaction with anti-light chains, but only in two of them a MGUS was detected (Figure 7a-f).  Immunohistochemistry with anti-TTR, anti-kappa light chains and anti-lambda light chains was performed in 35 out of 50 Congo red positive biopsies, and a binding with anti-TTR antibodies was observed in all cases (Figure 6a-c). Interestingly, four cases also showed a cross-reaction with anti-light chains, but only in two of them a MGUS was detected (Figure 7a-f). Congo red staining was positive in 50 of 69 cases (72.5%) (Figure 5a,b). Comparing the two most common mutations in our cohort, Congo red positive deposits were found in 21 of 24 Val30Met cases versus 13 of 27 Phe64Leu cases (p = 0.0035). We did not find any significant association between Congo red positive deposits and the clinical severity of polyneuropathy assessed by PND score, neither in the entire cohort (PND 1 and 2: 26 Congo red positive and 11 Congo red negative vs. PND 3A and 3B: 24 Congo red positive and 8 Congo red negative), nor considering the two most frequent mutations (Val30Met PND 2: 9 Congo red positive and 3 Congo red negative vs. Val30Met PND 3A and 3B: 12 Congo red positive and 0 Congo red negative; Phe64Leu PND 2: 6 Congo red positive and 6 Congo red negative vs. Phe64Leu PND 3A and 3B: 7 Congo red positive and 8 Congo red negative). Immunohistochemistry with anti-TTR, anti-kappa light chains and anti-lambda light chains was performed in 35 out of 50 Congo red positive biopsies, and a binding with anti-TTR antibodies was observed in all cases (Figure 6a-c). Interestingly, four cases also showed a cross-reaction with anti-light chains, but only in two of them a MGUS was detected (Figure 7a-f).  Electron microscope examination was available for 34 biopsies. In all of these cases, unmyelinated fibers were also reduced, and collagen pockets were observed (Figure 8d). Amyloid fibrils were detected in 11 cases (Figure 8a-c); 10 of these cases also showed amyloid deposits at Congo red staining while, in the remaining one with a Phe64Leu mutation, Congo red positive deposits were not found. Electron microscope examination was available for 34 biopsies. In all of these cases, unmyelinated fibers were also reduced, and collagen pockets were observed (Figure 8d). Amyloid fibrils were detected in 11 cases (Figure 8a-c); 10 of these cases also showed amyloid deposits at Congo red staining while, in the remaining one with a Phe64Leu mutation, Congo red positive deposits were not found.   Electron microscope examination was available for 34 biopsies. In all of these cases, unmyelinated fibers were also reduced, and collagen pockets were observed (Figure 8d). Amyloid fibrils were detected in 11 cases (Figure 8a-c); 10 of these cases also showed amyloid deposits at Congo red staining while, in the remaining one with a Phe64Leu mutation, Congo red positive deposits were not found.  Abdominal fat needle aspiration was available for 20 out of 69 (29%) patients, and it yielded positive results for Congo red staining in 8 of them (8/20, 40%). Two patients with infiltrative cardiomyopathy underwent a myocardial biopsy, which proved positive for Congo red staining.
Detailed demographic, clinical and pathological findings of the entire hATTR cohort and of the most frequently detected mutations are summarized in Table 1.

Discussion
Diagnosis of hATTR is challenging, and frequently misdiagnosis results in delayed therapy [4,21]. For many years, pathological evidence of amyloid deposition had been the gold standard for diagnosis, but in the last few years, also considering the introduction of new possible therapeutic approaches, genetic testing has become the first option after a clinical suspicion.
However, nerve biopsy remains a privileged instrument for a possible diagnosis in atypical cases in which a different clinical diagnosis was suspected, or to speculate about the mechanism of nerve damage induced by amyloid fibrils [8].
We examined a wide sample of nerve biopsies from hATTR patients coming from different Italian referral centers. The demographic characteristics of our cohort reflect those of classical late-onset hATTR, with a mean age of onset over 60 years and a male prevalence. We observed few early-onset cases, namely those carrying the Glu89Gln variant, the most frequent mutation in Sicily [17].
Regarding observed mutations, we confirmed that Val30Met was not the most common variant in Italy, with Phe64Leu being the most frequently found one in this cohort [5,6]. Although our data were focused on hATTR nerve biopsies, results were similar to those of the ATTRv amyloidosis Italian Registry [6], with the absence of Ile68Leu that classically shows a cardiological phenotype without nerve involvement.
The clinical spectrum included all the possible manifestations of hATTR. Hypertrophic cardiomyopathy was common in Val30Met, while autonomic dysfunction was frequent in Phe64Leu.
Regarding the polyneuropathy observed in this cohort, an axonal neuropathy was obviously the most common feature; it was found in 80.6% of patients and perfectly reflected the axonal loss showed by nerve biopsies, which ranged from moderate to complete loss in 94.2% of biopsies. In 16.4% of patients, a mixed polyneuropathy was described. In these cases, demyelinating findings were probably secondary to axonal loss, rather than caused by direct damage to the myelin, as frequently observed in other neuropathies [22,23]. However, since TTR is expressed in Schwann cells, a direct toxic effect on these cells, and on the nearby dorsal root ganglia, cannot be excluded [24,25]. There is progressive evidence that the oligomers of amyloidogenic proteins play a key role in mediating toxicity in other common neurodegenerative diseases, including Alzheimer's disease and Parkinson's disease [26]. Hence, biochemical stresses may be responsible for Schwann cell damage in patients with hATTR, in addition to the mechanical stress resulting from the formation of amyloid fibrils.
Interestingly, in two patients, a demyelinating neuropathy, according to EFNS/PNS criteria [27], was reported. In such cases, carrying the Val30Met and Thy78Phe mutations, respectively, a mechanic demyelination due to compression by amyloid deposition on nerve fibers [11,28] or a superimposed inflammatory process, as described in other cases of inherited neuropathies, could be hypothesized [25,[29][30][31]. Supporting both hypotheses, the nerve biopsy of the patient with Tyr78Phe showed a moderate axonal loss with inhomogeneous distribution among fascicles and Congo red positive amyloid deposits, while the cerebrospinal fluid revealed a moderate protein increase (56 mg/dL). Conversely, the nerve biopsy of the patient with Val30Met showed a moderate axonal loss with many fibers surrounded by thin myelin sheaths and Congo red positive amyloid deposits, while the cerebrospinal fluid revealed a normal protein level (30 mg/dL).
Myelin abnormalities were observed in ten biopsies. Unfortunately, cerebrospinal fluid examination was available only for the patients previously described. However, in one of these biopsies, inflammatory infiltrates were also found, and we observed inflammatory infiltrates in a further three biopsies, supporting the hypothesis of a superimposed inflammation [29,30]. Furthermore, an increase of protein in the cerebrospinal fluid may be found in hATTR [21], and it was present in about 60% of the exams available for our cohort. However, we cannot exclude that a protein increase in the cerebrospinal fluid examination is simply a consequence of radicular damage more than the sign of a superimposed inflammatory process. Uncommon pathological findings with myelin abnormalities and the presence of hyperproteinorrachia may be responsible for the frequent misdiagnosis of inflammatory neuropathies [21].
Regarding the sensitivity of nerve biopsies to detect amyloid, Congo red positive deposits were found in 72.5% of our biopsies. This percentage was higher if compared with abdominal fat needle aspiration, which was positive only in 40% of cases in our cohort. However, we should consider that this test was performed in 20 patients only. Regarding the different mutations, previous studies demonstrated that Phe64Leu mutation is not always associated with pathological deposits of amyloid in nerve biopsies or abdominal fat needle aspiration [4,15]. Analysis of our cohort confirms that Congo red positive deposits are less common with Phe64Leu mutations than with Val30Met mutations. The Congo red positivity of 87% with the Val30Met mutation, on the other hand, was similar to that reported in a recently published paper [32]. In contrast to this study, we did not find any significant correlation between the presence of Congo red deposits and the clinical severity of polyneuropathy, as assessed by PND score [32]. The proportion of early-onset patients in our cohort might explain this difference.
Immunohistochemistry with anti-TTR antibodies was able to confirm the nature of the deposits in all cases with Congo red positive deposits. However, we observed a cross-reaction with anti-lambda light chains in four biopsies. Two cases did not show MGUS at serum immunofixation. A possible cross-reaction among antibodies for amyloid typing has been described [11], and it confirms the importance of excluding hATTR through genetic testing when amyloid light-chain (AL) amyloidosis is suspected. On the other hand, in the other two cases, a MGUS was present, and a concomitant AL was diagnosed, confirming the importance of typing amyloid deposits [12].