The onset of MNGIE disease is usually between the first and second and decade of life, with an average age of onset at 18.5 years (Nishino et al., 2001; Garone et al., 2011); however, reported age of onset may not be accurate due to delay in diagnosis stemming from the subtlety of non-specific symptoms (Hirano et al., 1998). A few cases of late onset beyond the third decade and as late as the fifth decade have been reported, which were associated with compound heterozygous TYMP mutations and a less severe phenotype characterized by a partial reduction of thymidine phosphorylase activity (Marti et al., 2005; Massa et al., 2009). The earliest reported age of onset is five months of age (Garone et al., 2011). However, in a majority of patients, the first insidious symptoms manifest during childhood (Garone et al., 2011).

Skeletal muscle biopsy may show ragged-red fibers (due to abnormal proliferation of mitochondria in response to defective oxidative phosphorylation), ultra-structurally abnormal mitochondria, and abnormalities of both mtDNA and mitochondrial electron transport chain enzymes activities on enzyme analysis (Papadimitriou et al., 1998). However, it is important to note that ragged-red fibers are not always seen in MNGIE, as some patients do not display this histological abnormality (Szigeti et al., 2004b; Cardaioli et al., 2010).

Rectal biopsies show eosinophilic cytoplasmic inclusions in the submucosal ganglion cells (Perez-Atayde et al., 1998). Duodenal biopsies show focal muscle atrophy or absence, with increased nerve numbers, serosal granulomas and focal loss of Auerbach's plexus with fibrosis (Teitelbaum et al., 2002). Also, mtDNA depletion, mitochondrial proliferation and smooth cell atrophy are observed in the external layer of the muscularis propria in the stomach and small intestine (Giordano et al., 2006). Loss of interstitial cells of Cajal in the small bowel has also been reported (Zimmer et al., 2009; Yadak et al., 2018a).

Histopathological studies post mortem have failed to identify demyelination, neuronal loss or glial scarring in the areas of the brain white matter affected, as visualized by MRI (Szigeti et al., 2004a; Gramegna et al., 2018). However, the presence of albumin in the cytoplasm of reactive astrocytes was observed suggesting functional blood brain barrier alterations and consequent vasogenic oedema as a cause of leukoencephalopathy (Szigeti et al., 2004a). Furthermore, a mild perivascular gliosis was also observed in immunohistochemical analyses (Gramegna et al., 2018).

Ultra-structurally, peripheral nerve fibers show demyelination, and abnormal mitochondrial in Schwann cells (Hirano et al., 1994; Bedlack et al., 2004; Said et al., 2005). In addition to loss of myelinated fibers, nerve biopsies demonstrate mild perineural thickening, segmental demyelination, variation in internodal length and evidence of axonal regeneration (Hirano et al., 1994).

The rarity of MNGIE and its multisystem nature contribute to a complex clinical picture that is often difficult for non-specialist healthcare professionals to decipher and provide an early diagnosis (Filosto et al., 2011). This can lead to diagnostic delays of between 5 and 10 years (Lara et al., 2007; Taanman et al., 2009). Although confirmation of the diagnosis by testing for thymidine and deoxyuridine in the urine and plasma, combined with Sanger sequencing of the TYMP gene is straightforward, initial identification of this rare condition often requires a clinical interdisciplinary approach, leading to diagnostic delays, and unnecessary invasive diagnostic procedures, such as exploratory surgeries for gastrointestinal disturbance or unnecessary treatments, such as intravenous immunoglobulin before the diagnosis is made. A late diagnosis is often associated with a worse prognosis (Scarpelli et al., 2012; Çoban et al., 2013). This situation advocates the urgent need for the early diagnosis of MNGIE. Thus, thymidine phosphorylase deficiency should be suspected in cases where gastrointestinal and neurological involvement coexist, particularly where there is leukoencephalopathy on MRI or abnormalities of ocular motility (Scarpelli et al., 2012). The symptoms of MNGIE often resemble other conditions which are usually included in the differential diagnosis. Frequently, patients are incorrectly diagnosed with anorexia nervosa, inflammatory bowel disease, Crohn's disease, Whipple disease, chronic intestinal pseudo-obstruction, coeliac disease, chronic inflammatory demyelinating polyneuropathy and demyelinating forms of Charcot-Marie-Tooth disease (Said et al., 2005; Needham et al., 2007; Filosto et al., 2011; Garone et al., 2011; Demaria et al., 2016; Imperatore et al., 2017; Nagata and Buckelew, 2017; Kucerová et al., 2018). Phenotypes resembling MNGIE may be seen in patients with other mitochondrial DNA depletion syndromes including POLG or RRM2B mutations and Kearns-Sayre syndrome. These are often referred to as pseudo-MNGIE manifestations (Shaibani et al., 2009; Garone et al., 2011; Prasun and Koeberl, 2014). More recently two cases of MNGIE-like patients exhibiting POLG mutations were reported to manifest leukoencephalopathy and demyelinating peripheral neuropathy, which are characteristic not typically observed with these mutations (Yasuda et al., 2018). Another case study, reports two patients with a MNGIE-like phenotype exhibiting optic atrophy associated with a novel POLG mutation affecting the C- terminal sub-domain of the protein (Felhi et al., 2018).

There is a paucity of specific in vitro models of MNGIE described in the literature. A majority of the cell types implemented to date have no relevance to the organ systems affected in MNGIE and were used mainly to understand the effect of deoxyribonucleoside pool imbalances on cellular functions (Rampazzo et al., 2000, 2004; Pontarin et al., 2003). The first model developed was by Spinazzola et al. (2002), where fibroblasts derived from healthy controls and patients with MNGIE were used to study the contribution of thymidine phosphorylase in the deoxyribonucleoside pool imbalances. They examined the culture medium of cultured fibroblasts to determine the ability of healthy control cells and MNGIE patient cells to metabolize thymidine; in contrast to healthy cells, where a decline in media thymidine concentrations was measured, MNGIE fibroblasts were not able to catabolize thymidine, resulting in an increase in culture medium thymidine levels (Spinazzola et al., 2002).

Nishigaki et al. (2003) employed MNGIE-derived fibroblast to further evaluate the role of dysfunctional thymidine phosphorylase in the accumulation of deoxyribonucleoside pools (both thymidine and deoxyuridine) and with consequent mtDNA damage. Using patient-derived cell lines, 36 mtDNA point mutations, a TT to AA substitution and a single nucleotide deletion were identified. In MNGIE fibroblast cultures, cyclooxygenase activity was decreased, whereas deoxyuridine levels were markedly elevated. Also, an elevation in reactive oxygen species was observed, which was proposed to be a contributing factor to the accumulation of mtDNA point mutations. MtDNA sequencing of cultured fibroblasts and post mortem biopsies of skeletal muscle cells revealed a higher level of mtDNA point mutations in fibroblasts, whereas multiple mutations and deletions were observed at low levels in skeletal muscles. This suggests that fibroblasts primarily depend on anaerobic glycolysis rather than oxidative phosphorylation and therefore the absence of pressures on defective respiratory chain complexes in mitochondria results in the accumulation of nucleotide pools that generates a higher number of point mutations (Nishigaki et al., 2003).

In 2003, Song et al. used HeLa cells to show that increases in thymidine levels lead to an imbalance in dNTP pools, which ultimately result in mtDNA mutations. HeLa cells were cultured in medium supplemented with 50 μM thymidine. After 4 h of growth in thymidine-supplemented medium, the mitochondrial deoxythymidine triphosphate (dTTP) and deoxyguanosine triphosphate (dGTP) pools were shown to expand, whereas the deoxycytidine triphosphate (dCTP) pool dropped significantly, and the dATP pool dropped slightly. In whole cell extracts, the dTTP and dGTP pools also expanded, the dCTP pool decreased by approximately 50%, and the dATP pool remained unchanged. These changes in mitochondrial dNTP pools are consistent with a mutagenic mechanism involving the T-G mispairing followed by a next-nucleotide effect involving T insertion opposite to A. Supplementation of HeLa cells for 8 months with 50 μM thymidine, resulted in several mtDNA deletions (Song et al., 2003). It is noteworthy that to recreate MNGIE metabolite accumulations, the study implemented a 2.5-fold higher thymidine concentration to that observed in MNGIE patients, which is typically 20 μM thymidine (Song et al., 2003; Ferraro et al., 2005).

In 2005, Ferraro et al. conducted a similar experiment to Spinazzola et al. (2002), but used healthy skin and lung quiescent fibroblasts to demonstrate that mtDNA depletions are associated with post-mitotic cells. The study identified that mitochondrial deoxynucleotides are synthesized by two independent salvage pathways. In cycling cells, thymidine is salvaged by cytosolic thymidine kinase 1 (TK1) whereas in quiescent cells, thymidine is phosphorylated via thymidine kinase 2 (TK2) in the mitochondria, and the thymidine diphosphates then exported to the cytosol. Both cytosolic and mitochondrial thymidine phosphates undergo rapid turnover via deoxythymidine monophosphate (dTMP)/ thymidine substrate cycles. Therefore, quiescent cells lacking de novo synthesis and TK1 create a bias in dTTP pools, and this is further exacerbated in MNGIE where thymidine phosphorylase is lacking. Ferraro et al. (2005) cultured quiescent fibroblasts in medium supplemented with 10–40 μM thymidine and observed intra-cytosolic and intra-mitochondrial increase in dTTP and uridine triphosphates, both contributing to mtDNA depletions, concluding that mitochondrial DNA damage in MNGIE is predominant in post-mitotic cells (Ferraro et al., 2005).

González-Vioque et al. (2011), made use of murine liver mitochondria to show that mtDNA depletions are a consequence of a limited dNTP availability rather than a dNTP imbalance itself. The study demonstrated that excess of thymidine results in an increase of dTTP concentrations in mitochondria due to TK2 activity, with consequent secondary depletion of dCTP. TK2 phosphorylates both thymidine and cytidine competitively; each deoxynucleotide modulates the enzyme to consequently inhibit the phosphorylation of the other, although thymidine is more efficient at inhibiting the phosphorylation of cytidine. The addition of dCTP or deoxycytidine restored mtDNA depletions even in the presence of thymidine overload, confirming that mtDNA depletions are the result of a limited availability of substrates for mtDNA replication, caused by nucleotide depletion consequent to nucleotide overload rather than thymidine excess alone (González-Vioque et al., 2011).

Overall, in vitro models developed so far, have been informative and relevant for the understanding of the underlying biochemical and molecular mechanisms, associated with the deoxyribonucleoside pool imbalances and mtDNA depletions. However, to date tissue-specific models using cells relevant to the CNS, PNS and enteric system have not been developed and thus the study of alternative implications for the lack of thymidine phosphorylase expression on other biological pathways, including nervous tissue development and maintenance, has not been fully addressed.

Our research group has developed for the first time a MNGIE iPSC line which was used to generate a cerebral organoid model for the study of the CNS pathomolecular mechanisms and provide elucidations on the leukoencephalopathy observed (Pacitti, 2018; Pacitti and Bax, in press).

Murine models have proved to be very efficacious in the study of MNGIE, though it is important to consider the significant biological and hence metabolic differences between rodents and humans. This is exemplified by the metabolism of thymidine in the mouse, which is not only phosphorylated by thymidine phosphorylase, but also by uridine phosphorylase 1 and uridine phosphorylase 2; in the human, thymidine is solely metabolized by thymidine phosphorylase (el Kouni et al., 1993). To address this, Haraguchi et al. (2002) established a murine model based on double knock-out of Tymp^−/−/Upp1^−/− genes, whilst uridine phosphorylase 2 is not knocked-out in this model. Although this model recapitulates some features of the disease, it also displays some incongruences with the clinical scenario. The knock-out animals have a 10-fold increase in plasma thymidine and deoxyuridine, compared to >100-fold increase in the human. The mice also show cerebral oedema with hyperintense T2 MRI regions and axonal myelin fiber dilation without demyelination, however no peripheral neurological abnormalities were observed (Haraguchi et al., 2002). Also, Haraguchi et al. (2002) did not detect any mtDNA abnormalities in brain and muscle tissues of mice, suggesting that the loss of function of thymidine phosphorylase alone is not sufficient to cause MNGIE in this model. This led to the hypothesis that the adjacent gene SCO2 overlapping with TYMP sequences may be also contributing to the disease. The lack of mtDNA depletion in mice may be the result of a difference in mtDNA repair and replication, by which an increase in thymidine concentration may not affect the mitochondria of mice as it does in humans (Haraguchi et al., 2002).

A second double knock-out murine model of MNGIE was created in 2009 by Lopez et al. to characterize the biochemical, genetic and histological features of MNGIE in mice, and translate findings into the clinical picture (López et al., 2009). The resulting mice displayed undetectable thymidine phosphorylase activity in all tissues except in the liver, where the residual 17% activity was attributed to the expression of uridine phosphorylase 2. Mice displayed a 4- to 65-fold increase in thymidine levels in all tissues, with partial mtDNA depletions. Similarly, to the model developed by Haraguchi et al. (2002), the rodents manifested cerebral oedema with hyperintense T2 MRI signals in white matter, and late-onset cerebral and cerebellar white matter vacuoles without demyelination or axonal loss. However, in contrast to MNGIE patients, the model displayed mtDNA depletion, respiratory chain defects and histological abnormalities only in the brains, without any gastrointestinal or skeletal muscle involvement. López et al. (2009) suggested that the selective cerebral involvement observed in mice is possibly due to a number of factors, including differences in the life-span between species, as mice may not live long enough to accumulate sufficient mtDNA damage in most tissues or because the deoxyribonucleoside imbalance in humans is substantially more dramatic than in mutant mice. A third explanation is that high expression of TK2 in quiescent neuronal cells of rodent brains may contribute to an increased TTP production, thereby accelerating mtDNA damage in nervous tissues (Rylova et al., 2007). With regard to the contrasting findings of mtDNA depletions between Haraguchi's and Lopez's model, this could be explained by limitations in the analytical methods used for evaluating mtDNA aberrations (López et al., 2009).

In 2014, Garcia-Diaz et al. conducted a study to confirm the hypotheses generated by López et al. (2009) with regard to the role of thymidine accumulation in the pathogenesis of MNGIE, and in particular in the gastrointestinal involvement. López et al. (2009) failed to recapitulate the gastrointestinal dysmotility in mutant mice, and only replicated certain pathological features in mouse brains. It was speculated that the mild phenotype observed in the model is attributable to the short life-span of the animals combined with the modest increase in deoxyribonucleoside accumulation produced by mutant mice (which was 45-fold lower than that observed in MNGIE patients). Thus, to overcome this, Garcia-Diaz et al. (2014) supplemented mutant mice with exogenous thymidine and deoxyuridine to recreate a similar disproportion of deoxyribonucleoside concentrations as observed in humans, recapitulating the >100-fold increase in thymidine concentrations. The prolonged supplementation of deoxyribonucleosides in mutant mice resulted in the acquisition of biochemical abnormalities in the brain and small intestine, including mitochondrial DNA depletion as evidenced by cyclooxygenase deficiency observed through histological evaluations. Overall, treating double knock-out mice with thymidine was sufficient to enhance the phenotype of the model to recapitulate the clinical features of MNGIE, including weight loss, small intestine muscularis propria pathology, muscle weakness, leukoencephalopathy and decreased survival (Garcia-Diaz et al., 2014). However, in contrast to patients with MNGIE, who have multiple mtDNA deletions in brain, muscles, kidney and liver, the brain and muscle of 24-month old treated and untreated wild-type and Tymp^−/−/Upp1^−/− mice demonstrated similar levels of deleted mtDNA, suggesting that this is most likely due to aging rather than thymidine phosphorylase deficiency (Garcia-Diaz et al., 2014). Differences in the deoxyribonucleoside metabolism between humans and mice indicates the inadequacy of this model in recapitulating the human disease (Haraguchi et al., 2002; López et al., 2009; Garcia-Diaz et al., 2014).

Currently, there are no specific therapies for patients with MNGIE whose effectiveness has been evidenced in clinical trial studies. The current disease management guidelines aims to treat the specific symptoms that are evident in each individual and invariably requires the co-ordinated effort of different clinical specialities. Abdominal pain and nausea/vomiting secondary to gastrointestinal dysmotility are almost invariable, with patients treated symptomatically with analgesics, bowel motility stimulant drugs, anti-emetics and antibiotics for intestinal bacterial overgrowth (Teitelbaum et al., 2002; Oztas et al., 2010). Domperidone may be administered to control the post-prandial emesis and nausea (Yavuz et al., 2007). The reduction of epigastric pain episodes, especially in patients that are refractory to opiate pain management, can be achieved by performing a celiac plexus block with bupivacaine or by the selective blockade of the splanchnic nerve (Teitelbaum et al., 2002; Celebi et al., 2006). Pain may also occur in the limbs due to peripheral polyneuropathy and this can be treated with centrally acting agents such as amitriptyline, gabapentin and pregabalin (Hafez et al., 2014; Finsterer and Frank, 2017). Patients with MNGIE have an increased incidence of perforation of the gut, which generally requires emergency abdominal surgery (Granero Castro et al., 2010).

Malnutrition is a major problem in the majority of patients; various forms of parenteral nutrition, including total parenteral nutrition, are frequently required, but do not modify outcome (Wang et al., 2015). Complications of long-term parenteral nutrition use include the development of hepatic steatosis and cholestasis, and triglyceride hyperlipidemia. For patients with MNGIE there is the risk of metabolic oversupply from the lipid and carbohydrate components of the parenteral nutrition, leading to further mitochondrial toxicity. In later stages of the disease, patients are often unable to tolerate nasogastric nutrition due to gastrointestinal dysmotility (Wang et al., 2015). Portal hypertension may occur and be complicated by ascites and esophageal varices (Moran et al., 2008). These conditions are treated in the same way as when they occur in other conditions. Drugs that interfere with mitochondrial function should be avoided and hepatically metabolized drugs should be administered with care or contraindicated depending on the patient's liver function (Halter et al., 2010). Physiotherapy and occupational therapy input is usually required, particularly to address the neurological aspects of the condition.

A number of experimental therapeutic approaches are currently under investigation, including haemodialysis and peritoneal dialysis (Spinazzola et al., 2002; la Marca et al., 2006; Yavuz et al., 2007), allogeneic haematopoietic stem cell transplantation (AHSCT) (Hirano et al., 2006; Halter et al., 2010; Filosto et al., 2012), platelet transfusion (Lara et al., 2006), orthotopic liver transplant (OLT) (De Giorgio et al., 2016) and enzyme replacement (Bax et al., 2013). The therapeutic strategy common to all these approaches is to reduce or eliminate the pathological concentrations of thymidine and deoxyuridine, thereby ameliorating intracellular deoxyribonucleoside imbalances and preventing further damage to mtDNA, thus translating into clinical stabilization or improvement.

Plasma concentrations of thymidine were shown to be transiently lowered by haemodialysis, and infusions of platelets, which contain thymidine phosphorylase, were shown to reduce circulating levels of thymidine and deoxyuridine in two patients (Lara et al., 2006; Röeben et al., 2017). Disadvantages of these approaches are that haemodialysis is a burdensome procedure and long-term platelet therapy carries risks of developing immune reactions and transmission of viral infections and the short duration of effect.

AHSCT offers the possibility of a permanent correction of the thymidine phosphorylase deficiency but is limited by the availability of a matched donor. Patients are often in a poor clinical condition with an impaired capacity to tolerate transplant related problems and the aggressive conditioning and immunosuppressive chemotherapy (Halter et al., 2010, 2015). AHSCT also presents pharmacological challenges in terms of administering drugs with possible mitochondrial toxicity, and the requirement for parenteral administration due to disturbed gastrointestinal function and impairment of absorption. A published consensus proposal for standardizing an approach to AHSCT in patients with MNGIE recommended a recruitment restriction to patients in a stable clinical condition without irreversible end stage disease and having optimal donor (Halter et al., 2010). AHSCT is associated with an elevated mortality risk due to host-vs. -graft reactions and hospital acquired infections caused by the aggressive immunosuppressive regimen, combined with the disease (Halter et al., 2010; Filosto et al., 2012; Peedikayil et al., 2015). Halter et al. (2015) reported a mortality of 62.5% after the follow-up of 24 patients who received AHSCT (Halter et al., 2015). Patients who are oligosymptomatic are often reluctant to undergo AHSCT due to its high morbidity and mortality risk. A recent study suggested that the effect of AHSCT may be transient (Baker et al., 2017). Furthermore, a study focusing on the neuromuscular pathology of the small intestine in MNGIE highlighted that AHSCT may be insufficient to restore integrity of the enteric neurons and glia, thus without any short-term impact on the neurogenic and myogenic intestinal changes observed in later stages of MNGIE (Yadak et al., 2018a).

Due to the elevated expression of thymidine phosphorylase in the liver, solid organ transplantation is considered an alternative long-term therapeutic option (Boschetti et al., 2014). A case study has shown that OLT was able to normalize metabolite levels and provide mild improvements of neurological symptoms (De Giorgio et al., 2016; D'Angelo et al., 2017). The extent to which tissue damage can be reversed through the clearance of deoxyribonucleoside imbalances post- OLT has yet to be determined (De Giorgio et al., 2016).

Enzyme replacement therapy using autologous erythrocyte-encapsulated thymidine phosphorylase (EETP) is under investigation and has Orphan drug Designation by the FDA and EMA. The rationale for the development of EETP is based on thymidine and deoxyuridine being able to freely diffuse across the erythrocyte membrane via nucleoside transporters into the cell where the encapsulated enzyme catalyses their metabolism to the normal products (Figure 6). The products are then free to exit the cell into the blood plasma where they are further metabolized as normal. EETP is directed at ameliorating thymidine and deoxyuridine levels to slow the progression of MNGIE and stabilize the clinical condition and could therefore increase the chance of eligibility for AHSCT or OLT once a match is identified. Encapsulation of enzyme within the erythrocyte has the pharmacological advantages of prolonging the circulatory half-life of the enzyme and potentially minimizing immunogenic reactions which are frequently observed in enzyme replacement therapies administered by the conventional route. To date five patients have received EETP under a compassionate use programme, where clinical and metabolic improvements were observed (Moran et al., 2008; Halter et al., 2010; Godfrin and Bax, 2012; Godfrin et al., 2012; Bax et al., 2013).

Promising gene therapies for MNGIE are also under experimentation in murine models, using adenoviral vectors (AVV) targeting the liver for the correction of TYMP mutations for the restoration of normalized nucleoside metabolism (Torres-Torronteras et al., 2014). More recently, pre-clinical investigations of hematopoietic stem cell gene therapy in murine models have been conducted (Torres-Torronteras et al., 2016; Yadak et al., 2018a,b). A timeline of all investigational therapeutic approaches is summarized in Figure 7.

The development of drugs for rare diseases is confounded by a number of challenges such as small patient populations, phenotypic heterogeneity, incomplete knowledge of the disease pathophysiology or natural history and an absence of prior clinical studies. Consequently, the selection of clinical efficacy endpoints, which assess the way a patient feels, functions, or survives, can be an arduous process, particularly as validated endpoints appropriate for the disease are often unavailable.

There are generally no accepted endpoints for clinical studies in patients with MNGIE. This ultra-rare disease presents with usually a combination of cachexia, gastrointestinal dysfunction, and neuromuscular dysfunction. The determinants of morbidity and mortality in patients with MNGIE cannot be easily ascertained and owing to the rarity of the disease, there is no authoritative literature on the topic. The available case series of patients with MNGIE are small and with limited follow-up; the heterogeneity of the sources further limits the possibility to collate this information objectively. Additionally, there are no established patient reported outcomes specific to MNGIE. The experimental treatments for MNGIE aim to reverse the biochemical imbalances by eliminating the elevated systemic concentrations of thymidine and deoxyuridine. However, these metabolites do not provide objective measurements correlated to clinical status and are thus not suitable as end-points for predicting clinical benefit of therapeutic strategies. Several patients reported outcomes are available for specific symptoms or groups of symptoms (e.g., gastrointestinal, neuropathic) which are highly prevalent in patients with MNGIE; however, the extent to which those measurement instruments would be applicable to patients with MNGIE is unknown.

Despite genotypic differences and a variable phenotype, gastrointestinal symptoms including early satiety, nausea, dysphagia, gastroesophageal reflux, postprandial emesis, episodic abdominal pain, episodic abdominal distention, and diarrhea are cardinal manifestations of MNGIE and severely compromise nutritional homeostasis in almost all patients, leading to weight loss and cachexia. One of the largest case series available reports a “thin” body habitus in all patients, with weight loss from diagnosis averaging 15.2 kg (range: 5.9–30.0 kg) (Nishino et al., 1999).

Although clinicians treating patients with MNGIE, unanimously agree that weight loss is the key feature of the disease and has a major impact on their functional status, individual weight loss trajectories are not typically available in published case series. The consensus in personal communications with clinicians who treat those patients suggests that patients with MNGIE relentlessly lose weight and that this has a major impact on their functional status. Anecdotal evidence based on a review of case series and case studies in the literature suggests that organ failure, hepatic and gastrointestinal complications associated to cachexia are frequent causes of death in patients with MNGIE (Nishino et al., 2000; Garone et al., 2011).

The collection of uniform observational data through the operation of patient registries is one approach that is employed to identify suitable efficacy endpoints. Registries are particularly relevant to the field of rare diseases where the disorder has a heterogeneous presentation and information on the natural history is scarce. Of relevance to patients with MNGIE is the Rare Disease Clinical Research Network Natural History Study of MNGIE (NCT01694953) and also the North American Mitochondrial Disease Consortium which is currently collecting medical and family history, diagnostic test results, and prospective medical information; information from this will be invaluable for supporting the evaluation of new treatment modalities.

The Regulatory agencies are now recognizing the need for flexibility in the review of therapies for rare diseases and may consider approving a therapy based on a surrogate endpoint or biomarker as these can provide better objective measures of clinical benefit. Based on the identification of a number of dysregulated miRNAs in the serum of patients with MNGIE compared to age and sex matched healthy controls, Levene et al. are examining the application of a miRNA panel as a surrogate end-point biomarker in parallel with a clinical trial of EETP (Levene et al., in press).

St George's, University of London holds a licencing agreement with Orphan Technologies for the development of an enzyme replacement therapy for MNGIE. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.