Clinical Cues for the Early Diagnosis of Transthyretin-Related Polyneuropathy

Abstract

Background and Purpose

The estimated prevalence of hereditary transthyretin-related familial amyloid polyneuropathy (TTR-FAP) and the small number of known patients in Germany indicate that many patients with TTR-FAP remain undiagnosed, and may instead be classified as “idiopathic.” The aim of this study was to identify biomarkers for detecting TTR-FAP among a cohort of patients with idiopathic polyneuropathy (PNP).

Methods

Clinical evaluations (including the Neuropathy Impairment Score and Neuropathy Disability Score), nerve conduction studies (NCSs), quantitative sensory testing, and autonomic function tests were performed on 23 patients with TTR-FAP and 89 with idiopathic PNP. Discriminant analysis was then performed to identify variables useful for predicting TTR-FAP.

Results

Patients with TTR-FAP had paresis of the finger and thumb muscles, and reduced vibration perception and increased pressure pain in the upper and lower extremities. The NCSs showed that action potentials were smaller in the median, ulnar (both motor and sensory), and sural nerves in TTR-FAP. The sensory nerve conduction velocity was also reduced in the ulnar nerve. Autonomic neuropathy was confirmed by reduced sympathetic skin responses in the hands and feet in TTR-FAP. Multivariate discriminant analysis revealed that finger abduction strength, sensory ulnar nerve action potential amplitude, and vibration detection and pressure pain thresholds in the upper extremities were sufficient to correctly identify TTR-FAP in 81.3% of cases.

Conclusions

Detailed clinical and neurophysiological investigations of standard parameters in the upper limb may help to identify the otherwise-rare TTR-FAP.

Graphical Abstract

INTRODUCTION

More than 150 transthyretin (TTR) mutations have been described that cause hereditary transthyretin amyloidosis (ATTRv).¹ These mutations cause amyloid fibrils to accumulate and further cause dysfunction in various organs and tissues, particularly the peripheral nerves, leading to transthyretin-related familial amyloid polyneuropathy (TTR-FAP). Polyneuropathy (PNP) is often the first symptom, with the age at onset depending on the mutation and origin. The typical clinical presentation at TTR-FAP onset is distal symmetric PNP with neuropathic pain,² which is clinically similar to other neuropathies that are much more common. If left untreated, progressive organ dysfunction causes death within around 10 years, whereas early treatment increases life expectancy.³ Current causal treatments for TTR-FAP include liver transplantation, amyloid-stabilizing agents (tafamidis), and gene therapeutics (patisiran for RNA interference, and inotersen for RNA oligonucleotides). All therapies are most effective when initiated at an early disease stage,⁴ highlighting that early diagnosis is critical for a good outcome. However, misdiagnosis of TTR-FAP is common due to its rarity and initial nonspecific symptoms, and its detection can be delayed for years, which has a large negative impact on the course due to the lack of ability to replace damaged neurons.³ In line with that assumption of frequent misdiagnosis, the prevalence of TTR mutations responsible for TTR-FAP is estimated at 1:100,000 in Europe; this would correspond to about 800 cases in Germany. However, only 120 patients with symptomatic FAP are known.⁵ This discrepancy might be even stronger in countries with a higher prevalence of amyloid mutations, such as Sweden, France, and Italy. Portugal is the only country with a strong awareness of TTR-FAP.

In an unselected cohort, 30% of PNP cases are often classified as “idiopathic” after common causes are excluded.⁶,⁷ TTR-FAP cases will be found among these because genetic diagnoses for ATTRv and histological evidence of amyloid fibrils are not routine diagnostic tests. TTR-FAP might only be considered when PNP continues to progress steadily. However, the reasons mentioned above indicate that this leads to a significantly worse treatment outcome. There is therefore an urgent need for simple clinical examinations that provide clues to TTR-FAP. For this reason, we determined phenotypes for the two cohorts of TTR-FAP and idiopathic PNP and succeeded in distinguishing them with very few routine tests. These findings will help to improve patient care.

METHODS

Patients

This retrospective study included 23 patients with TTR-FAP and 89 with idiopathic PNP.

Patients with TTR-FAP were recruited during 2016–2021. The inclusion criteria for these patients was TTR mutation confirmed by Coutinho disease stage I (PNP-disability [PND] score: I and II)⁸ and genetic testing, and the presence of clinical PNP symptoms. Two of the cases had de novo amyloidosis after receiving a TTR-liver via domino liver transplantation. The age of these patients was 56.8±3.3 years (mean±standard error of mean). The TTR mutation Val30Met presented in 14 of the 23 patients (61%), and the remaining patients had the following mutations (1 in each patient): Ala50, Glu89Lys, Glu526Val, Thr49Ala, Leu58His, Val20 Ile, Ile88Leu, Val107, and Glu54Gln.

Patients with idiopathic PNP were tested for common causes and included if HbA1c was <5.6%, vitamin B12 was within the normal range, and presented with vasculitis markers including ANA, ENA, c- and p-ANCA, and with an immune electrophoresis in serum, and with nonpathological urine. Patients were excluded if they had a history of alcohol abuse (drinking >20 g/day for males and >10 g/day for females), current or previous tumors, or currently receiving or previously received treatment using neurotoxic substances. The age of these patients was 62.8±1.2 years.

The study was approved by the Ethics Committee of the Rhineland-Palatinate Medical Association (Landesärztekammer Rheinland-Pfalz; approval number: 2019-14182), and written informed consent was obtained from all patients who participated.

Clinical examinations

Patients were examined by experienced physicians (F.E.L. and P.B.). The Neurological Impairment Score (NIS) was calculated for the cranial nerves, thorax, and all limbs (NIS: 0–244 points).⁹ NIS subscores were assessed for the upper limbs (NIS-UL: 0–156 points) and lower limbs (NIS-LL: 0–88 points), and the motor items of the NIS were summed separately to determine the whole-body muscle status (0–192 points). Muscle strength was graded at 0–5 on the Medical Research Council (MRC) grading scale and compared among each muscle group. The neuropathy disability score (NDS) was also calculated, which includes measures of pain sensitivity, temperature, and vibration sense at the big toe, and the presence of the Achilles tendon reflex (0–10 points). Vibration sense was also measured at the thumb.

Questionnaires

All patients were asked to complete the painDETECT¹⁰ and Neuropathy Symptoms Score (NSS) questionnaires to assess their neuropathic pain symptoms. painDETECT consists of questions to estimate pain intensity, duration, patterns (persistent pain or attacks of pain), and quality (burning, tingling or prickling sensations, numbness, and temperature and pressure hyperalgesia). The NSS assesses unsteadiness in walking, numbness, burning sensations, and dysesthesia.

Technical investigations

Neurophysiological tests were selected that evaluate the function of small somatic and autonomic fibers and large nerve fibers.

Quantitative sensory testing

Quantitative sensory testing (QST) was performed on the right foot and hand dorsum according to the protocol of the German Research Network on Neuropathic Pain.¹¹ Thermal detection and thermal pain thresholds were considered in our study, as were the vibration detection threshold (VDT) and pressure pain threshold (PPT).

Autonomic function tests

The sympathetic skin response (SSR) reflects sweat gland activation in response to arousal.¹² This was recorded using electrodes on the hand and foot dorsum and the corresponding sites of the palm and sole. The peak-to-peak amplitude was assessed. The SSR amplitude approximately reflects the integrity of sudomotor nerve fibers. The SSR latency was not analyzed because sudomotor fibers are very slowly to conduct, and there is no pathological correlation such as fiber loss to slow it further.¹³ The definition of a normal range for SSR amplitude is disputable due to its large physiological variability. SSR is therefore considered “normal” if it can be reliably elicited (amplitude >0 µV).¹⁴

Cardiac autonomic neuropathy was assessed using the heart-rate variability. Seven statistical values in the time domain were obtained at rest, during deep respiration, and during the Valsalva maneuver.¹⁵

Nerve conduction studies

Motor (tibial, peroneal, median, and ulnar) and sensory (sural, median, and ulnar) nerve conduction velocities (NCVs) were measured on the right side of the body according to our laboratory standards.¹⁶ Compound muscle action potential (CMAP) amplitudes were measured as peak-to-peak values, and sensory nerve action potential (SNAP) ones were measured as baseline-to-peak values. Reference data were obtained from our laboratory.¹⁶

Statistical analyses

Data were transformed into Z-scores to compare parameters regardless of their physical units.¹⁷ Absolute reference data adjusted for age and sex were used to normalize the test results of each patient: Z-score=(single value_{test area}−mean_reference)/standard deviation_reference. Z>0 indicated “increased function” (relevant for QST) compared with the controls, whereas Z<0 indicate loss of function. Mann–Whitney U tests were performed in a sensitivity analysis without correction for multiple comparisons. Z-transformed group data (for TTR-FAP and idiopathic PNP) were compared with the reference data using two-sided independent t-tests.¹⁸ The chi-squared test was performed for categorical variables. Probability values of p<0.05 were considered significant. Missing values were not considered. Stepwise discriminant analysis (method of minimizing Wilks’ lambda) was performed to identify discriminators between TTR-FAP and idiopathic PNP. The stepwise forward method of minimizing Wilks’ lambda excludes variables with small discriminative power differences or strong correlations. Classification was achieved by calculating Fisher’s canonical linear discriminant function (D).¹⁹ D<0 classified patients as TTR-FAP, while D>1 classified them as idiopathic.

RESULTS

Patient characteristics

Patients with TTR-FAP presented with PND stage I (21.8%, n=5) or II (78.2%, n=18); all could walk without assistance, but 56.5% (n=13) reported pain. A family history of PNP was known to be present in 56.5% (n=13). When considering TTR-FAP subgroups Val30Met and non-Val30Met, we observed that both cohorts presented with similar clinical compound scores, but those without Val30Met had more often a walking disability (PND stage II, p=0.03) (Table 1). The results of the technical investigations (QST, nerve conduction study, and autonomic tests) did not differ significantly between the two TTR-FAP subgroups.

Table 1. Clinical and demographic characteristics of patients with TTR-FAP with and without Val30Met.

Characteristic		Val30Met (n=14)	Without Val30Met (n=9)	p
Age (yr)		54.9±4.8	59.8±3.9	0.44
		52 (25–79)	63 (40–76)
Male		57	67	0.65
PND stage				0.03*
	I	93	55
	II	7	45
NIS total		20.7±5.0	21.1±7.4	0.96
NIS-UL		4.3±1.5	3.9±1.6	0.85
NIS-LL		16.4±4.1	17.1±6.0	0.91
NIS-LL (muscle weakness)		6.1±2.9	7.0±4.1	0.86
NIS-LL (reflexes)		4.6±0.7	2.5±0.9	0.09
NIS-LL (sensory)		5.0±0.9	4.0±1.2	0.52
NDS		6.5±0.7	4.9±1.2	0.26

Data are median (range), mean±SEM, or percentage values.

^*p<0.05.

LL, lower limb; NDS, neuropathy disability score; NIS, Neurological Impairment Score; PND, polyneuropathy-disability; SEM, standard error of mean; TTR-FAP, transthyretin-related familial amyloid polyneuropathy; UL, upper limb.

Motor impairment was uncommon in patients with idiopathic PNP (10/23 with TTR-FAP vs. 3/89 with idiopathic PNP). Symptom duration at the time of assessment was longer in the idiopathic than the TTR-FAP group (69.3±8.2 month and 14.7±5.4 months, respectively; p<0.001). The pain and neuropathic symptoms were not significantly different in the two groups. Table 2 presents more details of these results.

Table 2. Clinical and demographic characteristics of the two study cohorts.

Characteristic	TTR-FAP (n=23)	Idiopathic (n=89)	p
Age (yr)	56.8±3.3	62.8±1.2	0.11
	56 (28–79)	62 (41–86)
Male	61	51	0.27
NIS total	20.8±4.1	13.7±1.0	0.10
NIS-UL	5.1±1.1	2.0±0.5	0.24
NIS-LL	15.7±3.3	11.0±0.9	0.12
NIS-LL (muscle weakness)	10.8±2.3	1.5±0.4	0.05
NIS-LL (reflexes)	3.4±0.7	4.2±0.3	0.61
NIS-LL (sensory)	2.6±0.6	3.9±0.3	0.90
NDS	5.9±0.6	5.7±0.3	0.91

Data are median (range), mean±SEM, or percentage values.

LL, lower limb; NDS, neuropathy disability score; NIS, Neurological Impairment Score; SEM, standard error of mean; TTR-FAP, transthyretin-related familial amyloid polyneuropathy; UL, upper limb.

Neurological examinations

We found that more patients with TTR-FAP had moderate motor symptoms (number of patients with muscle weakness: 8/23 [35%] for TTR-FAP vs. 9/89 [10%] for idiopathic) in the upper limbs with moderate weakness (MRC grade 4 or 5): hand flexors (p=0.002), finger abductors (p<0.001), short finger flexors (p<0.001), and finger extensors (p=0.006). Thumb abduction, adduction, and opposition were also weaker in TTR-FAP (p=0.012, p=0.005, and p=0.023, respectively). Vibration sense in the upper limbs was impaired more in patients with TTR-FAP (6.2±1.1 vs. 7.3±0.5, p=0.039).

Compound scores (NIS and its subscales, and NDS) (Table 1) and pain scores did not differ between the two groups.

Quantitative sensory testing

Temperature detection and pain thresholds were pathological in the back of the hand and foot dorsum in both cohorts, with no intergroup differences. However, we found differences in PPT in the upper limbs (632±285 and 962±464 in TTR-FAP and idiopathic patients, respectively; p=0.006) and lower limbs (753±356 vs. 1037±405, p=0.005), and in VDT in the upper limbs (5.8±1.0 vs. 7.3±4.9, p=0.039) (Fig. 1).

Fig. 1. Results of the pain pressure (A) and vibration (B) tests in the upper limbs. Patients with TTR-FAP were more sensitive to pain pressure in the upper and lower limbs. Vibration perception at the wrist was impaired more in patients with TTR-FAP. ^*p<0.05; ^**p<0.01. Idiop, idiopathic; LL, lower limb; PPT, pressure pain threshold; TTR-FAP, transthyretin-related familial amyloid polyneuropathy; UL, upper limb.

Nerve conduction

Patients with TTR-FAP presented with significantly smaller potentials in the ulnar and median nerves, with CMAPs in the median nerve of 7.7±4.1 mV and 10.4±5.7 mV (p=0.031) in TTR-FAP and idiopathic patients, respectively; CMAPs in the ulnar nerve of 9.7±4.7 mV and 12.09±3.89 mV (p=0.011); SNAPs in the median nerve of 10.8±9.9 mV and 27.5±26.6 mV (p<0.001); SNAPs in the ulnar nerve of 15.1±16.0 mV and 26.6±26.6 mV (p=0.004); and SNAPs in the sural nerve of 5.5±7.2 mV and 10.3±12.7 mV (p=0.029) (Fig. 2).

Fig. 2. Differences in action potential amplitudes in nerve conductions studies. Patients with TTR-FAP presented with smaller motor nerve potentials in the median and ulnar nerves. Sensory nerve potentials were also smaller in the median, ulnar, and sural nerves, which is a sign of further axonal damage in this group. ^*p<0.05; ^**p<0.01; ^***p<0.001. Idiop.PNP, idiopathic polyneuropathy; m, motor nerve potential; NCS, nerve conduction study; s, sensory nerve potential; TTR-FAP, transthyretin-related familial amyloid polyneuropathy.

NCV only differed in the sensory median nerve (37.1±24.6 mV and 53.8±26.6 mV in TTR-FAP and idiopathic patients, respectively; p=0.005).

Autonomic function tests

Patients with TTR-FAP had significantly smaller SSR amplitudes in the lower limbs (85±29 mV vs. 475±95 mV, p<0.0001) and upper limbs (257±112 mV vs. 677±105 mV, p<0.001) (Fig. 3). There were no significant differences in the other parameters.

Fig. 3. SSR results in both the upper and lower limbs. Autonomic dysfunction in the TTR-FAP group was evident from lower SSR amplitudes in the upper and lower extremities. ^***p<0.001; ^****p<0.0001. Idiop, idiopathic; LL, lower limb; SSR, sympathetic skin response; TTR-FAP, transthyretin-related familial amyloid polyneuropathy; UL, upper limb.

Multivariate discriminant analysis

The prerequisites for variables to be included in the stepwise discriminant analysis (minimize Wilks’ lambda) were 1) strong scientific justification, 2) differences between TTR-FAP and idiopathic PNP in univariate tests, and 3) a reasonable amount of valid data to avoid losing too many cases. The following variables were included: SSR in the upper and lower extremities, finger flexor and thumb adduction, amplitude in the ulnar motor and sensory nerves, amplitude in the motor median nerve, VDT in the upper extremities, and PPT in the both extremities. NCV in the ulnar sensory nerve and cold detection threshold in the upper limb were also included, because a previous retrospective analysis found that they could be useful for discrimination.²⁰ Cases with missing values were excluded from this test, leaving 20 patients with TTR-FAP and 76 with idiopathic PNP being included in the discriminant analysis.

Stepwise discriminant analysis selected finger spread strength, VDT, and PPT in the upper limbs, and SNAP in the ulnar nerve as discriminant variables that were sufficient to distinguish TTR-FAP from idiopathic PNP (Wilks’ lambda=0.69). The remaining variables did not contribute to the discrimination ability. Fisher’s linear discriminant function was as follows:

D=-14.8+2.1×(finger spreaders)+0.473×(VDT)+0.001×(PPT)+0.016×(sensory ulnar amplitude).

This formula correctly classified patients in 81.3% of cases (75.0% and 82.9% of TTR-FAP and idiopathic patients, respectively) (Table 3). The formula exhibited a sensitivity of 54% (range=33.9%–75.5%) and a specificity of 93% (range=83.7%–97.6%).

Table 3. Results of the discriminant analysis.

Polyneuropathy		Predicted group		Total
		TTR-FAP	Idiopathic
Original
	TTR-FAP	15 (75.0)	5 (25.0)	20 (100)
	Idiopathic	13 (17.1)	63 (82.9)	76 (100)

Data are n (%) values. The discriminant formula correctly classified patients in 81.3% of cases (75.0% and 82.9% of TTR-FAP and idiopathic patients, respectively).

TTR-FAP, transthyretin-related familial amyloid polyneuropathy.

DISCUSSION

Routine evaluations of distal symmetric PNP do not include screening for rare diseases such as TTR-FAP.⁷,²¹ Early stages of TTR-FAP might therefore escape diagnosis if they do not receive special attention. However, early diagnosis is essential because of the causal treatments that could be offered to the patients to prevent progression but not reverse symptoms. The present study found compelling evidence that clinical routine investigations, particularly those assessing nerve function in the upper extremities, could fill this gap and highlight cues for genetic or histological screening for amyloidosis. The multivariate analysis that we applied after standard statistical approaches provided a simple formula that discriminates TTR-FAP from idiopathic PNP using only four parameters: intrinsic hand muscle weakness, vibration sense and pressure pain in the thumbs, and ulnar SNAP on routine NCV testing. This discriminant formula was able to correctly classify four of five cases.

The identified discriminative variables might mostly relate to the upper extremities due to pathological findings in the lower extremities being frequent in most distal symmetric neuropathies, resulting in fiber loss patterns in the feet or legs not being useful for discriminating between etiologies. Furthermore, upper-limb impairment in TTR-FAP might be due to—unlike most neuropathies—TTR-FAP not being a pure distally dying-back axonopathy,²² and that amyloid deposits are patchy and located along the entire length of the nerve²³,²⁴ and dorsal root ganglia.²⁵

Most cases could be allocated into groups even though the TTR-FAP group was genetically heterogeneous, but we found that four patients with a Val30Met mutation and one with Glu74Gln had been misclassified. This was likely due to the patients all having a family history and being diagnosed and treated very early, which prevented PNP progression.

Isolated test results often do not differ sufficiently in these two groups to allow a differential diagnosis on their own, and so a univariate analysis may not have been sufficient. We therefore performed a multivariate data analysis.

Discriminant analyses have been widely used in recent bioinformatics analyses of cohort studies²⁶ and allow the identification of predictors for group allocation, especially if they are based on metric-independent variables such as the Z-scores used in our study.²⁷ Fisher’s nonstandardized linear discriminant function then allows group allocation for individual patients based on independent variables in the analysis.

This result was consistent with our previous finding that TTR-FAP is mostly differentiated from other neuropathies such as diabetic PNP and chronic inflammatory demyelinating neuropathy by clinical signs in the upper extremities.²⁰ However, the control group of “idiopathic” neuropathies in the present study was the more-appropriate clinical control group because testing for amyloidosis is not routine during a PNP workup, as mentioned above. Our findings indicated that patients with TTR-FAP present signs of hyperalgesia and hyperpathia (higher pressure sensitivity) more often as indicators of small-fiber damage and central sensitization, and experience more damage to myelinated nerve fibers (greater impairment of vibration perception, decreased strength, and smaller nerve action potentials). Previous studies have indicated that patients with early-onset TTR-FAP often present with damage to the small nerve fibers, while those with late onset may present with other patterns, with sensory or sensorimotor symptoms that start at the feet, distally at all four extremities or only at the upper extremities.²⁸ Cases with late onset and early onset were equally distributed among our patients. Our study also corroborated the precocious motor involvement and greater axonal loss especially in the upper limbs (e.g., in a neuronopathy) in TTR-FAP, which was consistent with a previous study.²⁹

One limitation of the present study was the smallness of the TTR-FAP cohort, which was due to the rarity of the disease in Germany. However, the results were consistent with those in the literature, increasing their credibility. The heterogeneity of the mutations included in our study may have been a source of the variability, as some variants are known to be more likely to induce PNP, and the rate of PNP progression may vary. In any case, all patients included in the study had PNP, and the clinical features of patients with the most common mutation (Val30Met) were similar to those of patients without Val30Met (except for walking disability). Furthermore, we considered both the early- and late-onset phenotypes in our cohort. Nevertheless, the ages of the patients in both groups did not differ significantly, improving the plausibility of the results. Finally, our results are only valid for distal symmetrical polyneuropathies and not for asymmetric or multiplex neuropathies. Such neuropathies often prompt nerve biopsies, which will often identify amyloid deposits.

Our results are clinically relevant since the discrimination tests were simple and can be assessed using a standard clinical exploration (strength and vibration), and measurements of nerve conduction (part of a standard PNP evaluation) and pain pressure (can be easily assessed using a pocket-size algometer). They can be implemented in guidelines for routine patient assessments. Although a single measurement is not sufficient to distinguish potential patients with TTR-FAP from idiopathic ones, the neurologist could identify early candidates for further more-invasive and costly investigations by paying attention to the four discriminant variables included in the formula. In this way, early detection of TTR-FAP can be improved to avoid delays in diagnosis and treatment.

Availability of Data and Material

The datasets generated or analyzed during the study are available from the corresponding author on reasonable request.