Creatine is a physiological compound that was first isolated by the French chemist Eugène Chevreul in the 1830s 1 . It is supplied to the body both via endogenous biosynthesis and the diet 2 3 4 . Its name (the Greek ‘kreas’ means meat) originates from its abundance in muscle in which its energetic role has previously been documented 5 6 7 8 . Creatine can be phosphorylated by creatine kinases, leading to the storage of a high-energy phosphate bond of ATP in the form of phosphocreatine 3 . Upon demand, and due to the reversible activity of creatine kinases which can transform ADP back into ATP, phosphocreatine can release the stored energy 3 . This creatine/phosphocreatine cycle represents a vital reserve of energy in tissues with high energetic needs such as muscles and the brain 3 . In recent years, the physiological role of brain creatine has been extended to include the modulatory control of neurotransmission which, in addition to guanidinoacetate (GAA) toxicity, has improved our understanding of brain function and its severe impairment in disorders affecting creatine biosynthesis and transport 9 10 11 12 .

The first step of creatine biosynthesis occurs essentially but not exclusively in the kidney and involves the transfer of an amidino group from arginine to glycine by the enzyme L-arginine/glycine amidinotransferase (AGAT) (EC 2.1.4.1) to form GAA and L-ornithine. Subsequently, N-guanidinoacetatemethyltransferase (GAMT) (EC 2.1.1.2) catalyzes the transfer of a methyl group from S-adenosyl methionine (SAM) on GAA. This second step occurs notably in the liver and leads to the formation of creatine and S-adenosylhomocysteine (SAH) 3 . Once synthesized, creatine may be released into the bloodstream. It can be internalized by cells via a specific plasma membrane Na⁺/Cl^- dependent creatine transporter (SLC6A8) and stored as phosphocreatine 3 . Finally, about 1.7% of the creatine/phosphocreatine pool undergoes spontaneous degradation each day forming creatinine that is excreted in urine 3 . Urinary excretion of creatinine is a function of body muscle mass and in practice is a useful marker of kidney function. Little or no creatine is taken up by the brain from the systemic bloodstream because the blood brain barrier is practically impermeable to creatine. The brain must therefore secure most of its requirements in creatine by endogenous synthesis 10 11 13 . The crucial role of SLC6A8, which has been reviewed recently 14 , provides ground for understanding how its deficiency results in a creatine collapse in brain.

Extracerebral and intracerebral routes for creatine supplies concur and their respective contributions differ in the immature (i.e. fetal and perinatal) and mature brain. In the immature brain, the extracerebral supply is favored due to a better efficacy of SLC6A8 in the blood–brain barrier (BBB). This means that extraction of creatine from the blood is predominant over intracerebral creatine biosynthesis which is physiologically limited by low GAMT activity in the immature brain 14 . SLC6A8 deficiency affects the immature brain creatine content through impairment of BBB uptake. In the mature brain, the contribution of BBB SLC6A8 becomes physiologically lower and intracerebral biosynthesis becomes the major pathway for the supply of brain creatine 14 . SLC6A8 deficiency therefore affects the mature brain creatine content through impairment of intracerebral creatine biosynthesis as a result of reduced SLC6A8-driven transfer of GAA from brain GAA to creatine-synthesizing cells. Differences existing between the extracellular/intracellular routes for the brain creatine supply explain why creatine supplementation is more efficient in two other creatine disorders, AGAT and GAMT deficiencies, when given at a presymptomatic stage of the disease when BBB uptake of creatine is physiologically optimal 14 15 . An illustrated account of creatine biosynthesis and transport is provided in the Additional file 1.

Primary creatine deficiency disorders (PCD) are a new class of inborn errors of metabolism to which belong AGAT (OMIM 602360) and GAMT (OMIM 601240) deficiencies, both inherited as an autosomal recessive trait, and SLC6A8 [CT1, CRTR] deficiency (OMIM 300036), an X-linked disorder 10 11 12 16 17 18 19 20 . The three disorders share a relative reduction of brain creatine and phosphocreatine levels 19 , and accordingly, the clinical presentation is predominated by neurological signs 10 11 12 16 17 18 19 20 . However, these disorders still remain under-diagnosed whether considered together 16 17 18 19 20 or individually (AGAT 21 22 23 24 25 , GAMT 26 27 or CRTR 28 29 30 31 32 33 deficiency).

In this paper we present the results of a large retrospective study of urinary screening for PCD on 6,353 patients in whom the etiology of the underlying neurological disease was unknown at the time of urinary analysis of creatine and its metabolites.

We retrospectively collected the metabolic results of urinary screening for PCD (GAA and creatine:creatinine ratios) from 6,353 patients with neurological disease of unknown origin (4,426 male patients and 1,927 female patients) hospitalized in six French university hospitals (Angers, Lille, Lyon and Paris (Hôpital Necker Enfants Malades, Hôpital Robert Debré and Hôpital Raymond Poincaré)) over a period of 28 months. Data with regard to GAA and creatine levels in plasma were also collected when available. For all patients with an abnormal result, a second sample was analyzed after a 24-hour diet devoid of meat and fish according to the recommendation of Arias et al. 34 . A patient was considered “at risk of PCD” (AGAT, GAMT or SLC6A8 deficiency) when the biochemical abnormality was confirmed on this second sample. Subsequently a molecular study and/or a functional test were performed to confirm the diagnosis. For the genetic study, informed consent was obtained from each patient or from their parents if probands were under 18. For patients with a confirmed diagnosis of PCD, extensive clinical data were collected, including familial, obstetrical and personal histories, main symptoms having led to consult, the specialty of the physician who requested metabolic investigations, the patient’s age at diagnosis, clinical signs present at diagnosis and results of ¹H-MRS exploration. All patients with other diagnosed conditions (in particular, urea cycle disorders and remethylation defects) were discarded from the study.

In the present study we have identified 16 new patients diagnosed with a PCD over a period of 28 months, as well as three affected siblings. Intellectual disability and attention deficit were, along with motor and speech delays, the main symptoms reported by the medical staff involved in the diagnosis of these disorders. Nevertheless, it is surprising that intellectual disability, which was a sign observed by the medical examiners in all patients with PCD, was not the main cause leading the parents to consult. This might reflect the difficulty of identifying intellectual disability in the very young child or in turn to of associating intellectual disability with PCD. These disorders are known to progress via irreversible lesions of the brain, so therapeutic measures are needed to be taken early in the life of the patients. For GAMT patients, supplementation with creatine together with a GAA lowering strategy (ornithine supplementation with or without arginine restriction) may represent an efficient treatment strategy 43 44 45 46 and may provide some medical neurological benefit for GAMT patients. In these conditions, patients may recover a normal brain creatine peak as detected by ¹H-MRS as well as normal urine GAA levels. They also improve clinically with better language and social development and apparently some regression of the epilepsy 47 . By contrast, in patients with SLC6A8 deficiency, supplementations with creatine and its precursors arginine or glycine, though improving muscular signs, were found to be without substantial benefit on cognitive and psychiatric signs, and failed to modify the creatine signal in ¹H-MRS imaging of the brain 48 , suggesting the inefficacy of these supplementations in improving brain as stated by other studies 4 19 30 49 50 51 52 . Though there is hope to treat efficiently this group of patients through the discovery of creatine pro-drugs capable of entering the brain and cells via a by-pass of the SCL6A8 creatine transporter, some clinical studies also question the inefficacy of the therapeutic measures mentioned above and, on the contrary, show beneficial effects with creatine and/or arginine supplementations in patients with SLC6A8 deficiency. In a heterozygous female patient with intractable epilepsy, treatment with creatine combined with arginine and glycine completely resolved seizures 53 . Creatine supplementation was also described to improve the neurological, language and behavioral status and was associated with a rise in the brain creatine peak as demonstrated by MRS in a child with SLC6A8 deficiency 54 . In a recent study, creatine deficient patients were also shown to be improved by a L-arginine-based therapy which positively impacted daily living skills, lowered the frequency of epileptic episodes and induced a mild increase in brain creatine and phosphocreatine MRS signals although normal cerebral levels of these metabolites were not recovered 55 .

The calculated prevalence of PCD in our cohort was 0.25%. This result was not expected because most studies report a higher prevalence of PCD, notably SLC6A8 deficiency, which is generally estimated between 1% and 3% of the population affected with intellectual disability 56 57 58 59 . However, this prevalence has been considered to be an overestimation of the real prevalence of PCD. Because PCDs are monogenic disorders, their prevalence was proposed to be no different from that of other nonsyndromic diseases such ARX (Aristaless-Related homeobox gene located on X chromosome) and therefore closer to 0.1% - 0.3% 60 . Thus, for the first time, our screening study provides strong practice-based evidence confirming this estimated prevalence for PCD. The girl diagnosed with SLC6A8 deficiency presented a severe phenotype similar to affected male patients, an observation that is not surprising in view of recent work 34 61 62 also describing this X-linked disorder in the female population with intellectual disability. In this respect, screening for SLC6A8 deficiency in female patients should be included in the diagnostic workup since this disorder still remains under-diagnosed. However, it should be noted that if the urinary U-CT/CTN ratio is not increased the diagnosis cannot be ruled out because it has been shown that the majority of female patients with a heterozygous mutation in SLC6A8 have a normal ratio, although the average ratio of this group is increased 62 . It might be recommended to perform brain MRS and/or SLC6A8 gene studies in female patients with suspected creatine transporter deficiency.

Five mutations were identified for GAMT gene in 6 unrelated families. These included 3 new mutations, c.289C > T, c.391 + 15G > T and c.577C > T, and two previously described mutations, c.299_311dup13 27 and c.506G > A 63 (Table 2). The new mutation c391 + 15G > T was considered pathogenic by the creation of an alternative splicing donor site in intron 3 (Alamut software), by the segregation of this mutation in his parents and because it was linked with decreased GAMT activity in cultured fibroblasts.

The SLC6A8 gene was analyzed in 11 patients from 9 unrelated families. Nine mutations were identified throughout the gene, including 4 new mutations (2 missense (c.1208C > A and c.926C > A), 1 frameshift (c.930delG) and 1 splicing (c.1393-1G > A) mutation), and 5 previously described mutations (c.321_323delCTT 64 , c.942_944delCTT 65 , c.1006_1008delAAC 56 , c.778-2A > G and c.1519_1543del 61 ) (Table 3). In accordance with the results of Clark et al. 56 , we found a fairly high proportion of frameshift and splicing mutations in our patients, and approximately one fifth of the mutations of SLC6A8, but not GAMT, were attributed to neo-mutation, germinal or somatic mosaicism events.

Interestingly, DNA studies performed in our patients clearly showed that mutations were essentially patient/family specific and highlighted the absence of hot spot mutations in these genes. Finally, as a mutation was identified in each patient for whom DNA was analyzed, the study of the target gene appears to be a key diagnostic step for PCD in suspect patients, keeping in mind that secondary creatine disorders such as deficiencies of ornithine delta-aminotransferase (EC 2.6.1.13) 66 67 , urea 68 69 70 71 and homocysteine/methionine 72 73 cycles, and succinate semialdehyde dehydrogenase (SSADH) (EC 1.2.1.24) 74 75 may also be alternative causes of abnormal laboratory values of creatine metabolism markers.

Screening for PCD in the population with neurological disorders of unknown etiology has shown that these patients express important clinical signs of PCD which mainly include intellectual disability, speech delay, epilepsy, attention deficit and autistic behavior.

In this respect, some improvement might be made by conducting an information campaign among general and specialized educational staff to contribute to better recognizing intellectual disability and hence earlier referral of the child for professional medical evaluation. Improved anamnesis in regard to the familial history of possible patients might be also helpful. As suggested in the present study and highlighted in other work, both female and male patients can be concerned by severe PCD. However, and importantly, our study has shown that the prevalence of PCD remains low in the patients with unexplained neurological symptoms. Finally, the mutational spectrum of PCD has been extended by the new mutations detected during the present screening effort.

When PCD was suspected informed consent was obtained from patients and/or their parents for the DNA analysis of GAMT or SLC6A8 genes.

The authors declare that they have no competing interests.

DC and MJCC contributed equally to this work and are first co-authors. Authors (DC, MJCC, GB, KMM, DD, JMC; AC, VV; JFB; JMP, GSI, OD, AA, LL, VD, MG, RD, NP, JV) working in the six major French university hospitals (including those of Lille, Lyon and Angers and the Hôpital Necker Enfants Malades, Hôpital Robert Debré and Hôpital Raymond Poincaré from the Public Assistance of Parisian Hospitals) collaborated to group and to generate clinical biological, biochemical and genetic data. The other authors were also involved in the clinical follow-up (AG, GP, BC, FR, KD, MT) or biological chemistry (SB) of some of the patients and in the gene studies (FC, FP, DH and GSA). DC, MJCC, GB, NP and JV analyzed and integrated the clinical and paraclinical data. JV coordinated the writing of the manuscript with DC, MJCC and GB. All authors read and approved the final manuscript.