A total of 12 patients (6 male, 6 female) were included in this cohort (Table 1). The median age at diagnosis was 56 years (range 3–69). All nine adult patients were symptomatic at diagnosis (median age at diagnosis 61 years, range 42–69); five adults were diagnosed after recognizing the typical brain MRI abnormalities, three adults were diagnosed by metabolic screening after MRI abnormalities gave rise to suspicion of a metabolic disorder and one adult was diagnosed after an older sibling was diagnosed with AMACR deficiency (patient 5). Three children were included (median age at diagnosis was 5 years, range 3–6). One child (patient 10) was diagnosed by metabolic screening because of the incidental finding of elevated liver enzymes in blood tests performed because of fatigue; one asymptomatic child (patient 12) was diagnosed following the diagnosis in an older sibling (patient 11), who underwent genetic analysis because of oculocutaneous albinism, which revealed a homozygous deletion of chromosome 5p13.3 encompassing five genes, including the AMACR gene [17]. The average time of follow-up was 6 years (range 1–9).

Most patients had homozygous missense variants in the AMACR gene (reference sequence NM_014324). Three patients were homozygous for the c.364 C > T (p.His122Tyr) variant, three patients were homozygous for the c.154T > C (p.Ser52Pro) variant and one patient homozygous for the c.557 A > G (p.Glu186Gly) variant. Two siblings had a deletion of the AMACR gene due to chromosome 5p13.3 deletion and three patient had compound heterozygous variants: c.154T > C (p.Ser52Pro) plus c.512G > A (p. Arg171His), c.154T > C (p.Ser52Pro) plus c.364 C > T (p.His122Tyr), and c.155 C > A (p.Ser52*) plus c.1040dup (p.Glu348Argfs*4). Only the c.154T > C (p.Ser52Pro) variant and chromosome 5p13.3 deletion have been previously reported [1, 17]. In all patients, AMACR deficiency was also confirmed by elevated plasma levels of pristanic acid and detection of C27- bile acid intermediates (R)-DHCA and (R)-THCA in blood.

All adult patients presented with neurological signs that led to further investigations (Table 2). The most common signs among adult patients were sensomotor axonal neuropathy (n = 6/9), ataxia (n = 4/9) or cognitive decline with or without behavioral problems (n = 4/9). All these patients were above 40 years of age at the time of diagnosis. Patient 8 presented at the age of 46 years with subacute onset of headache, cognitive impairment, aphasia, apraxia and the development of right sided hemianopia over the course of several days. MRI of the brain showed no signs of recent ischemia, but did show symmetrical T2 hyperintense signal in the thalami and brain stem. EEG showed diffuse slowing of the left hemisphere. Partial spontaneous recovery occurred, but residual complaints of mild aphasia and hemianopia were reported. No possible triggering factor like preceding fasting or weight loss was identified in this patient. Two patients (patient 4 and 7) had a history of focal seizures (age 17 and 46 years respectively, at occurrence), which responded well to antiepileptic drugs. One patient (patient 7) was known to have intellectual impairment and suffered from an episode of depression with psychotic features. All other patients had normal cognition before cognitive or behavioral problems occurred. Five out of nine adult patients were previously diagnosed with retinitis pigmentosa with a median age at diagnosis of 45 years (range 30–61), but most of them reported vision problems already years before. The median delay to diagnosis of AMACR deficiency after the diagnosis of retinitis pigmentosa was 24 years (range 0–33). No patients had suffered from episodes of rhabdomyolysis or encephalopathic episodes other than the above described stroke-like episode.

One patient was diagnosed with liver fibrosis (patient 3), while all patients underwent regular abdominal ultrasounds and in most cases FibroScan®. In patient 3, ultrasound and MRI of the liver showed multifocal tumors in both liver lobes. The ultrasound was performed as part of regular work-up shortly after he was diagnosed with AMACR deficiency at the age of 66 years. Biopsy showed hepatocellular carcinoma and he died six months later of liver failure. Another patient had a sibling who died of liver cancer, who was not evaluated for AMACR deficiency at the time. Two patients had a history of cholecystectomy due to gallstones (patient 4 and 6, respectively around age 40 and 15 years). No patients were reported to have suffered from neonatal cholestasis. Other malignancies reported were ocular melanoma (patient 6) and urothelial carcinoma (patient 9), the latter diagnosed during follow-up.

Most patients did not develop any new disease signs during follow-up (see Table 2). All three children (patients 10–12) were asymptomatic at diagnosis and did not develop any signs during follow-up of at least seven years.

Magnetic resonance imaging (MRI) of the brain was performed in all patients. Contrast-enhanced images were available in four patients. The three children who underwent brain MRI (patient 10–12) between ages of 7 and 8 years showed no T2 hyperintensities, nor any other abnormalities. The age at the first performed MRI in the adult patients ranged between 35 and 67 years.

In all adult patients, the first MRI already showed characteristic abnormalities. The typical MRI pattern consisted of symmetric T2 hyperintense signal abnormalities in the thalami, midbrain and pons (Fig. 2). The thalami were affected in eight of the nine patients. The midbrain was affected in all nine adult patients, including T2 hyperintensity of the substantia nigra in all patients and superior colliculi in seven patients. The grey matter of the pons was selectively involved in eight patients, sparing the corticospinal tracts as well as the ascending sensory tracts. This created either a trident-like configuration or a closed omega (ɷ) (Fig. 3), which was noted in four and two patients, respectively.

FLAIR images of patient 4 show a normal appearance of the medulla oblongata. The signal abnormalities of the pons are continuous with the affected substantia nigra. The red nucleus and thalamus are also symmetrically affected

Cropped FLAIR and T2-weighted images of three different patients (from left to right: patient 3, 9 and 5) are shown, centered on the pons. Typical pontine gray matter involvement can be appreciated on both sequences, despite variation in quality of the images. Notice the sparing of the corticospinal tracts and the ascending sensory tracts, among which the medial lemniscus

The medulla oblongata, middle cerebellar peduncles and cerebellar white matter showed no signal changes. Normally, the T2 signal of the dentate nucleus and the red nucleus is in general lower than the adjacent white matter in adults over 40 years of age. This was the case in none of our adult patients. The T2 signal of the dentate nucleus and red nucleus was evidently higher than white matter in three patients and five patients, respectively, and isointense with the surroundings in the other adult patients. The superior cerebellar peduncles exhibited subtly increased T2 signal in four patients.

No abnormal contrast enhancement and no diffusion restriction were observed. In four patients atrophy of the brain stem disproportionate for age was noted, particularly of the midbrain and superior cerebellar peduncles. In patient 8, who presented with a stroke-like episode, the MRI showed cortical FLAIR hyperintensity and subtle cortical swelling of the left cerebral hemisphere, and presumably reactive hyperintensity of the left pulvinar (Fig. 4), in addition to AMACR-related abnormalities in the brain stem and thalami. In three patients, multifocal T2 hyperintensities of presumed vascular origin were observed in the cerebral white matter (patient 2, 6 and 7 at age 66, 62 and 52 years, respectively). In another patient confluent presumably vascular T2 hyperintensities were found in the cerebral white matter together with lacunar infarcts, small vermis infarcts and a cortical infarct (patient 4 at age 66).

T2-weighted and FLAIR images of patient 8 are shown here. She presented with a (transient) stroke-like episode of aphasia, apraxia and hemianopsia, reflected by the cortical hyperintensity and swelling of the left cerebral hemisphere and reactive hyperintensity of the left pulvinar. The abnormalities in the pons are subtle and punctate in this patient, additionally the signal of the dentate nucleus is abnormal. The red nucleus and thalamus are involved

Follow-up MRI of the brain with an interval greater than three years was available in four adult patients and one child. Three of the adults showed unchanged MRI abnormalities. Patient 4 underwent MRI with 29-year interval (age 37 and 66) and showed new periventricular white matter abnormalities, and progression of T2 hyperintensities at follow-up. In the child with follow-up MRI after 5 years still no abnormalities were found. In five out of the nine adult patients, the typical MRI abnormalities eventually led to the diagnosis. However, in multiple cases there was a delay of several years before the patient was referred and the already present MRI abnormalities were recognized as fitting with the above-described pattern seen in AMACR deficiency.

Bile acid intermediates (R)-DHCA and (R)-THCA were elevated in all patients. The median DHCA level in plasma was 4.8 µmol/L (reference value 0.0 µmol/L) and THCA level in plasma was 5.6 µmol/L (reference value 0.0–0.1 µmol/L) (Fig. 5a). In one patient the primary bile acid cholic acid in plasma was below the lower limit of normal (reference value 0.1–4.7 µmol/L) and in five patients the primary bile acid chenodeoxycholic acid was below the lower limit of normal (reference value 0.7–10.0 µmol/L). In four patients, urinary bile acids were measured and C27-bile acid intermediates in urine were detectable in two of those patients. Pristanic acid was elevated in all patients (median 77.0 µmol/L, reference value 0.0–1.6 µmol/L). Phytanic acid was normal in six patients (median 9.5 µmol/L, reference value 0.49–9.88 µmol/L), but it was unclear whether some of these patients were already on a phytanic acid-restricted diet at the time of analysis.

A. Boxplot of (R)-dihydroxycholestenoic acid (DHCA) and (R)-trihydroxycholestenoic acid (THCA) in all patients (n = 12). The upper reference range for DHCA is 0.0 µmol/L and 0.1 µmol/L for THCA. The median is shown as a solid line. B. Boxplot of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) in 11 patients. The upper reference range (40 U/L) is indicated by the dotted horizontal line, the median as a solid line. Outliers are shown as black dots

Blood samples obtained after diagnosis, prior to possible vitamin supplementation, were analyzed for liver function tests, fat soluble vitamins and coagulation parameters. No coagulation parameters and vitamin levels were available for patient 1. Patient 3 was excluded from analysis of liver function tests because of liver cancer at the time of diagnosis. All 11 other patients had normal (total) bilirubin and alkaline phosphatase levels at the time of diagnosis. Only one adult had a slight elevated gamma-glutamyl transferase level in plasma. Five patients had elevated aspartate aminotransferase (AST) levels (median 51 U/L, range 46–146), three of whom are children. All three children also had elevated alanine aminotransferase (ALT) levels (54, 98 and 219 U/L respectively), which none of the adults had (Fig. 5b).

In six out of ten patients vitamin D deficiency was diagnosed, in three out of ten patients vitamin A deficiency and two out of ten patients vitamin E deficiency. Only one patient showed a slight prolonged prothrombin time (PT) and three patients showed a slight prolonged activated partial thromboplastin time (aPTT). Factor VII activity was measured in five patients, who all showed reduced activity (median 64%, reference value 80–140%), none of these patients showed clinical signs of a coagulopathy.