Wolcott-Rallison syndrome (WRS) was named after Drs Wolcott and Rallison, who first described this syndrome in three affected siblings [1]. It is also known as multiple epiphyseal dysplasia and early-onset diabetes mellitus, according to the main clinical manifestations recognized initially.

WRS is characterized by insulin-requiring diabetes that generally appears during the neonatal period or in the first six months of life, with a frequently acute presentation of severe diabetic ketoacidosis at disease onset [2–4]. Two patients with a later onset have been reported, at 14 months [4] and 30 months [2]. Hyperglycemia is initially the only manifestation of the disease, the rest of the biological assessment being normal. Anti-islet cell (ICA), anti-insulin, anti-glutamic acid decarboxylase (GAD) and anti-tyrosine phosphatase (IA2) antibodies are negative. With time, short stature and skeletal dysplasia with radiographic abnormalities progressively develop and are generally diagnosed after diabetes onset, although early signs, including delayed or difficult walking, osteoporosis, deficient mineralization or mild bone abnormalities may be present on close examination and radiography as early as the onset of diabetes. Additional clinical features, which vary between patients, are usually diagnosed after diabetes onset.

This is a rare disease, with fewer than 60 cases described in the literature [3, 4]. In the large majority of cases, affected individuals are from populations in which consanguineous marriages are frequent, such as the Middle-East, North Africa, Pakistan and Turkey. Despite its recognition as a very rare disease, with very few cases reported in the literature before the first genetic study leading to gene identification, WRS now appears as the most frequent cause of Permanent Neonatal Diabetes Mellitus (PNDM) in consanguineous families ([4] and C. Julier, unpublished data). It is likely that the syndrome was rarely diagnosed before the identification of the responsible gene, and may still be underdiagnosed, as patients may die before expressing the minimum clinical features of neonatal/early-onset diabetes and epiphyseal dysplasia or because the association may not be recognized in younger patients.

The two main clinical features are well described in the first publication of Wolcott and Rallison [1], and are represented by neonatal/early-onset diabetes and multiple epiphyseal dysplasia. Hepatic dysfunction, manifesting in the form of elevated hepatic enzymes, liver enlargement and recurrent acute liver failure, is the third most frequently observed manifestation, and now appears as a characteristic feature of this syndrome. This review is based on the clinical characteristics of patients in our own experience and on the description of cases reported in the literature, as well as two recent compilations of large series of WRS patients: a description of new cases and literature review of 38 patients from 23 families [3] and a recent study of 29 WRS patients from 25 families [4].

Bone dysplasia

WRS is characterized by multiple epiphyseo-metaphyseal dysplasia whose major features affect the long bones, pelvis and vertebrae, while the skull is usually normal. Dysplastic changes on knee radiographs are illustrated by enlarged and irregular metaphyses with prominent beaks. Femoral and tibial epiphyses are flattened. On the hand, the carpal bones and phalanges show tubulation defects and appear short and enlarged. The carpal centres are small and irregular, the proximal phalanges show especially dense and cone-shaped epiphyses (Figure 1A) [5]. Interestingly, bone mineralization is affected, as was noted in the early descriptions of the syndrome [6]. Osteopenia is associated with thin cortices and poorly calcified epiphyses which are undeveloped. Multiple and frequent fractures can be observed. Blood calcium and phosphorus values are normal. At the spine, the characteristics correspond to those of mild platyspondyly. Vertebrae show irregular upper and lower plates with frequent ossification defects at the anterior edge (Figure 1B). Changes are especially marked at the dorso-lumbar level with a consequent appearance of thoracic kyphosis and/or lumbar lordosis. Clinically, the chest is rather broad and flattened in the antero-posterior direction (dwarfism with short trunk) (Figure 2A). At the pelvic level, the radiographs usually show abnormal iliac wings and dysplastic acetabular roofs (Figure 1C). Dislocation or subluxation of the femoral heads is a common complication. Femoral epiphyses are flattened with coxa vara. Clinically, difficulty in walking or running is frequent, with a "duck-like" gait related to stiff hips and limited abduction. In some cases, the extreme skeletal condition may manifest with severe cervical spine instability, which may result in spinal cord compression and motor neuropathy in arms and legs (T. Barrett, pers. communication).

The clinical course of WRS is very variable, including within the same sibship. Early age at onset of diabetes, bone dysplasia and recurrent hepatic failures, which are the most characteristic features, also show variability between patients. Typically, onset of diabetes occurs before age 6 months and bone dysplasia is diagnosed before one or two years of age. Hepatic episodes may occur at any time during the course of the disease, and may be the first presenting manifestation after diabetes onset. However, age at onset has been reported up to 2.5 years [2] and one patient was described with no bone abnormalities at age 32 years [4]. The presence and severity of other manifestations, i.e. renal failure, exocrine pancreas deficiency, intellectual deficit, hypothyroidism, neutropenia and recurrent infections, are variable between patients. The variability in age at onset, severity and nature of the clinical manifestations and frequency of severe episodes of clinical manifestations explain the variable course of the disease between patients, with survival ranging from a few weeks to 35 years. In some patients, some of the clinical manifestations may be missing due to the young age of the patient, and death may occur before the full diagnosis of WRS is performed.

Histological studies performed on the bone, pancreas and liver of WRS subjects have shown very specific features, that are valuable to our understanding of disease mechanisms. In the first publications describing this syndrome, a defect in ossification was well documented on bone biopsy, with particular abnormalities of the cartilage in the form of hypertrophic chondrocytes [1, 34, 35]. Further studies reported an irregular proliferation of chondrocytes in the resting cartilage, with the presence of excessively thin ramified collagen fibers in the spongy bone [9].

In the liver, biopsies performed outside of the recurrent episodes of liver dysfunction typically show progressive fibrosis with mild steatosis and intrahepatic cholestasis [7]. Postmortem examination showed steatosis and lesions of massive necrosis, which is non-inflammatory and has a very remarkable aspect, where cells are isolated or grouped in clusters juxtaposed to normal hepatocytes, producing a mosaic appearance (Figure 5; S. Collardeau-Frachon, pers. communication). Similar necrotic lesions were also observed in the kidney (S. Collardeau-Frachon, pers. communication).

The pancreas is typically hypoplastic with a reduction of acinar tissue that exhibits evidence of necrosis, with similar histological characteristics to the liver (Figure 6; S. Collardeau-Frachon, pers. communication). Langerhans islets are smaller than normal, in relation with the severe β cell defects, as evidenced by immunohistochemical study showing a major reduction of insulin staining, whereas the majority of cells stain for glucagon (Figure 7) [36].

Overall, both in human WRS and Perk KO mice, the necrotic lesions observed in the liver, kidney and exocrine pancreas appear histologically very similar, contrasting with the absence of similar features in the endocrine pancreas and the bone. Indeed, in β cells and osteoblasts, detailed mouse studies showed clear evidence for proliferation and differentiation defects, rather than cell death [22, 24, 33]. In view of the broad range of mechanisms controlled by PERK, it is likely that the variety of cellular consequences of PERK deficiency observed in different organs and cell types reflect the variety of the mechanisms and pathways that may be affected, directly or indirectly. The consequences for β cells, resulting in diabetes in WRS patients, appear to affect principally the proliferation and differentiation of the cells, as well as proinsulin trafficking and insulin secretion [33]. Similar mechanisms acting on the highly secreting osteoblasts are likely responsible for the bone defect observed in WRS, as demonstrated in Perk KO mice [24]. In contrast, the necrosis appearance observed in the liver, kidney and exocrine pancreas suggests the implication of distinct mechanisms, timing or regulations, which may result in cell death in a cell-specific manner. Detailed investigation of the exocrine pancreas in exocrine-specific Perk KO mice showed that the mode of cell death in this tissue is oncosis rather than apoptosis, and there was no evidence of defective protein synthesis and secretion, a normal appearance of the ER in these cells, and no evidence of ER stress, but a strong inflammatory response [23]. Finally, recent studies in Perk KO mice suggest that PERK may act as a metabolic sensor in the β cells, to modulate the trafficking and quality control of proinsulin to the physiologic demand in insulin [33]. This adaptive mechanism is likely to play a central role in frequent forms of diabetes and metabolic diseases. The remarkably similar clinical and histological characteristics of human WRS and Perk KO mice stress the value of detailed studies of global and organ-specific Perk KO mice in order to dissect the variety of tissue-specific defects and understand disease mechanisms[22].

WRS should be differentiated from other forms of neonatal or early-onset insulin-dependent diabetes, based on the clinical presentation. Ultimately, EIF2AK3 mutation screening confirms (or excludes) the WRS diagnosis. The hypothesis of a variant form of WRS, not caused by EIF2AK3 mutation and with variant clinical features, remains a possibility, but should be confirmed in additional cases [2].

WRS is always permanent after disease onset, which easily distinguishes it from transient neonatal diabetes, which is caused by other genetic defects. Most cases of PNDM occurring before the age of 6 months are thought to be largely genetically distinct from typical T1D, based on the absence of distortion of HLA class II allele distribution compared to control individuals [37, 38]. In contrast to T1D, WRS patients do not have β cell specific autoantibody at disease onset.

Other genetic causes of PNDM have been identified, in which diabetes may be isolated (non-syndromic) or syndromic. Non-syndromic diabetes can be caused by dominant or recessive mutations in the potassium channel genes KCNJ11 and ABCC8, or the insulin gene (INS), or by recessive mutations in glucokinase gene (GCK); additional neuro-motor and neuro-psychological abnormalities are frequently observed in patients with KCNJ11 and ABCC8 mutations. In outbred Caucasian populations, heterozygous mutations in KCNJ11, ABCC8 or INS genes may account for almost 40% of PNDM, compared to less than 5% in consanguineous families; in contrast homozygous mutations at INS, GCK or ABCC8 genes may account for a total of 30% and EIF2AK3 mutations for almost 25% of PNDM patients in consanguineous families [4]. Although WRS is easily distinguished from these other forms of PNDM based of the additional clinical features, they may be mistaken at the time of diabetes diagnosis, when the rest of the syndrome is not yet expressed. Careful inspection of early signs of deficient bone mineralization and skeletal dysplasia should be searched for at the time of diagnosis. Although the age at onset of diabetes and birth weight may be additional distinctive features, in that patients with EIF2AK3 and ABCC8 homozygous mutations tend to have a later age at onset and slightly greater birth weight than patients with homozygous mutations at GCK and INS [4], this criteria may not be fully diagnostic because of the overlap of distributions.

Other rare autosomal recessive forms of PNDM, which are associated with various organ impairments, have been genetically characterized and are easily differentiated from WRS: PDX1 mutations cause pancreas agenesis with additional exocrine pancreas deficiency [39], or a variant form with only neonatal diabetes [40]; PTF1A mutations cause pancreas and cerebellar agenesis [41]; GLIS3 mutations cause neonatal diabetes, congenital hypothyroidism and other clinical features [42]. FOXP3 mutations (X-linked) are responsible for neonatal diabetes syndrome associated with X-linked immune dysregulation and enteropathy [43].

The short stature related to skeletal dysplasia, regardless of other dysfunctions that occur in patients with WRS, may be evocative of other spondylo-epiphyseal dysplasias such as mucopolysaccharidoses. Early-onset T1D occurring in these patients may also lead to a similar phenotype as WRS. In such cases however, disease onset is likely to be older, and β cell specific autoantibodies should differentiate them from true WRS.

In summary, in view of the straightforward genetic diagnosis, we recommend EIF2AK3 mutation screening for diagnosis in all neonatal/early-onset diabetes patients (at least before age of 6 months) with clinical features specific for WRS, or in the absence of extra-pancreatic manifestations in PNDM patients born from consanguineous parents.

Genetic counselling of WRS is strongly recommended. As a rare autosomal recessive disease, with no clinical manifestations in parents, genetic counselling can only be proposed to parents of a previous diagnosed WRS child, with confirmed EIF2AK3 mutation. Recurrence risk is 25% in siblings of WRS patients. In cases of extended consanguineous families with a genetically diagnosed WRS case, other related consanguineous couples should also benefit from genetic counselling, and their carrier status for the EIF2AK3 mutation determined.

In view of the early disease onset and severe condition and prognosis of WRS at present, prenatal diagnosis is strongly recommended for this disease. Pregnancies are generally uncomplicated, and the only prenatal diagnosis of WRS is genetic diagnosis, based on genotyping the EIF2AK3 mutation(s) that is(are) present in both parents (heterozygous carriers).

In terms of diabetes, close therapeutic monitoring should be recommended because of the tendency for frequent acute episodes of both hypoglycaemia and ketoacidosis. Under this context, treatment with insulin pump is recommended, especially in small infants, in order to optimize the prevention of severe hypoglycaemia. Generally, in contrast with patients affected with other forms of diabetes, treatment with insulin should not target a very tight control of blood glucose, in order to avoid hypoglycemia, which may trigger episodes of acute aggravation of the disease. In practice, children with WRS have a risk of developing acute multi-organ failure during intercurrent illness. It is very important to inform the patient's relatives in order that they can recognize these episodes very early so that hospitalization can be organized very rapidly and appropriate symptomatic treatment instituted. Bone fractures may be frequent and must be managed at the orthopaedic level. In WRS patients, interventions under general anaesthesia increase the risk of acute aggravation, as a consequence of the toxicity of anaesthetics, and should be avoided when possible. Similarly, any drug or vaccine not strictly necessary should be limited, taking into account the risk of triggering secondary liver and/or kidney failure.

The prognosis is poor, and WRS patients generally die at a young age. In a review of 19 patients with known age at death, only 3 died at 10 years or older [3]. Two patients were reported with a longer survival, until 35 years for one patient [2], and alive at 32 years old for one patient [4]. In general, in the final episode of aggravation, the patient is initially admitted to the hospital with flu-like symptoms and fever. Death occurs later in a situation of multi-organ failure with predominant liver and renal dysfunction, hepatic failure being sometimes associated with encephalopathy. Remarkably, acute liver episodes had not been reported in the patient who survived until age 35 years.

In view of the observed ER dysfunction in WRS patients, intervention strategies targeting a reduction of ER stress or other pathways involved in this dysfunction are of potential interest for therapy and possibly prevention. Recent studies targeting ER stress reduction using chemical chaperones, glucagon-like peptide-1 (GLP-1) agonists or other strategies have reported promising results in vitro and in vivo (cellular and animal models), and in particular regarding protection of β cells from ER stress [31, 44]. Wolfram syndrome (WFS), a monogenic diabetes, is caused by mutations in the WFS1 gene, another important ER gene. Interestingly, treatment of Wfs1 KO mice with pioglitazone was found to reduce β cell loss and protect mice from diabetes [45]. As ER stress related mechanisms, or possibly other ER disturbance, have been suggested not only for monogenic diabetes such as WRS and WFS but also in frequent forms of diabetes, such as T1D and T2D, and in other common conditions including obesity and cardiovascular diseases, such intervention strategies may have very large applications for health purpose.