Introduction

Maturity-onset diabetes of the young type 2 (MODY2) is a monogenic form of diabetes with autosomic dominant transmission caused by heterozygous, inactivating mutation in the glucokinase gene (GCK). Loss-of-function GCK mutations impair glucose-sensing of the pancreatic beta cells (and liver) that in turn increase the threshold of pancreatic glucose-stimulated insulin secretion [1]–[3]. As a result, MODY2 patients have a moderate, not-progressive increase in fasting glucose and HbA1c levels and impaired glucose tolerance at the oral glucose tolerance test. In addition, carriers of GCK mutations are usually not prone to micro- or macro-vascular complications and, with rare exceptions, do not need pharmacological intervention [4].

It is now clearly established that MODY2 patients under treatment with insulin or oral anti-diabetic drugs can discontinue therapy without deterioration of their metabolic control. Moreover, inappropriate insulin treatment can induce iatrogenic weight gain and onset of insulin resistance (D.I., personal observation, and others [5], [6]). In fact, the typical MODY2 patient requires less frequent clinical surveillance than patients with other forms of diabetes. As a consequence, to diagnose MODY 2, especially in the pediatric age, is of paramount importance to ensure the appropriate clinical management and save healthcare resources [7].

The exact prevalence of MODY of any type is unknown, probably because their phenotypes overlap with the classical forms of diabetes. Their overall prevalence is estimated at around 1% of all diabetic cases, with MODY2 ranging from 30% to 60% of all MODY sub-types depending on clinical setting and geographic origin [8], [9]. Various studies indicate that only a small proportion of MODY are correctly identified. For instance, in the UK, only 5% to 20% of all patients theoretically affected by MODY are predicted to be correctly diagnosed [9], [10]. The number of patients referred for MODY genetic testing differs widely among diabetic clinics, and more experienced clinicians refer more patients for testing while keeping a good detection rate [9].

Genetic testing is highly specific and sensitive and represents the gold standard for diagnosing MODY2. However, this test has the drawback of being still expensive and not easily accessible. In 2008, the European Molecular Genetics Quality Network (EMGQN) published the guidelines for the selection of patients for MODY molecular testing, thereby providing an important common ground for European geneticists, pediatricians and diabetologists [5]. The guidelines included most of the relevant clinical parameters that define MODY, whereas others, such as the assay of the autoantibodies responsible for classical type 1 diabetes, were not taken into account [11], [12]. In this context, the guidelines have a poor detection rate and diagnostic power that can be improved, particularly in pediatrics, by adding other clinical criteria [13], [14].

Not all physicians treating diabetic patients have the expertise to recognize the clinical features of genetic forms of diabetes. Consequently, they hesitate to propose molecular testing, and it is likely that many patients remain undiagnosed. In this scenario, we designed and validated a 7-item flowchart (7-iF) to identify patients that have a high probability of carrying GCK mutations, thereby improving the detection rate and minimizing the number of unnecessary tests. We chose to study this specific type of MODY because is the commonest in Italy, at least in the pediatric population [15], [16], and because its diagnosis provides a perceptible impact on both patient's quality of life (no need of treatment) and health care costs (less frequent follow up visits, no stick for glycemic controls or drugs to provide). The 7-iF consists of seven questions with binary answers (yes/no) about data easily obtainable during a standard clinical examination, and therefore it can be applied in all clinical settings. The 7-iF includes the most recent criteria for the etiological diagnosis of diabetes (i.e. autoimmune diabetes antibodies, HbA1c levels and familiarity) and is particularly addressed to general practitioners and physicians working outside specialized centers, who are more prone to overlook this diagnosis. Here we show that the 7-iF has high clinical utility in comparison to the clinical suspicion alone and that a wide application of this method might 1) help less experienced clinicians in suspecting MODY2 patients and 2) reducing the number of unnecessary tests.

Methods

Prospective evaluation of the 7-iF

We evaluated the clinical utility of the 7-iF in a “prospective cohort” consisting of 921 patients followed-up at the diabetic outpatient clinic of the Department of Pediatrics of the Second University of Naples (S.U.N.). The 7-iF was validated in two phases (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Validation of the 7-item flowchart.

https://doi.org/10.1371/journal.pone.0079933.t001

Retrospective validation

We validated the clinical utility of the 7-iF on 372 patients (“first retrospective cohort”) addressed to the genetic test on the basis of clinical suspicion and with a MODY2 diagnosis confirmed by molecular test. This part of the study was conducted by physicians of the Italian Study Group on Diabetes of the Italian Society of Pediatric Endocrinology and Diabetology (ISPED) (Figure 1). The “first retrospective cohort” consisted of MODY2 patients from all over Italy to reduce population stratification and selection bias. The cohort consisted of patients for whom complete medical records were available; a single “unknown” to any of the 7-iF questions was considered an exclusion criterion. The “second retrospective cohort”, including 210 patients, was composed by the patients referred to (1) the Department of Molecular Medicine and Medical Biotechnologies, University of Naples “Federico II”- CEINGE and (2) the Pediatrics Unit, IRCCS Casa Sollievo della Sofferenza Hospital, S. Giovanni Rotondo (FG), with a clinical suspicion of MODY2 and the indication to the genetic test. We collected these latter data for patients referred in five years' time, before the beginning of the present study.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. P.I. and city of each participating ISPED centers.

A. Fabrizio Barbetti (Roma), B. Corrado Mammì (Reggio Calabria), C. Maurizio Delvecchio (San Giovanni Rotondo), D₁. Nadia Tinto (Napoli), D₂. Enza Mozzillo (Napoli), E. Luigi Pianese (Ascoli Piceno), F. Sonia Toni (Firenze), G. Bruno Pasquino (Bolzano), H. Valeria Calcaterra (Pavia), I. Ivana Rabbone (Torino), J. Barbara Felappi (Brescia), K. Francesca Cardella (Palermo), L. Valentino Cherubini (Ancona), M. Franco Mammì (Locri), N. Anna Paola Frongia (Cagliari), O. Francesco Gallo (Brindisi), P. Stefano Tumini (Chieti), Q. Sonia Lucchesi (Livorno), R. Carla Maria Monciotti (Padova), S. Susanna Coccioli (Francavilla Fontana), T. Amedeo Vergerio († deceased) (Feltre), U. Stefano Zucchini (Bologna), V. Francesco Cadario (Novara), Riccardo Lera (Alessandria), X. Andrea Scaramuzza (Milano).

https://doi.org/10.1371/journal.pone.0079933.g001

Results

Retrospective validation of the 7-iF

We tested and validated the 7-iF on a large series of MODY2 cases, the “first retrospective cohort”, recruited, merely on the basis of the clinical suspicion, by the members of the Study Group on Diabetes of the ISPED that is constituted by Italian hospital and university clinics caring children with diabetes. All these patients received a molecular diagnosis of MODY2 before the beginning of the present study. Data from 19 Italian outpatients' clinics and 5 medical genetics laboratories were pooled and analyzed (Figure 1). Each patient was unequivocally identified by a unique code to avoid overlapping. We received the records of 372 patients. Figure 2 summarizes the results of this retrospective validation. The upper panel shows the percentage of patients meeting each item. Most patients fulfilled each criterion. Although satisfactory (around 80%), the single item with the lowest prediction value was HbA1c. The lower panel shows the percentage of MODY2 patients fulfilling the 7 items, 6 out of 7 items or less than 6 items. About 67% of the MODY2 patients studied fulfilled 7 criteria out of 7, 25% of them proved positive to 6 out of 7 criteria and only 8% were positive just to 5 or 4 item of the 7-iF. This result indicates that, if strictly applied on this cohort of patients, the 7-iF would have correctly addressed to the genetic test the 67% of these patients missing 33% of them.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Validation on the “first retrospective cohort”.

(A) Percentage of MODY2 patients positive to each item of the 7-iF. (B) Percentage of MODY2 patients positive to 7, 6 or less than 6 items of the 7-iF.

https://doi.org/10.1371/journal.pone.0079933.g002

In order to validate the precision of the 7-iF, we investigate a “second retrospective cohort” composed by 210 pediatric diabetic patients referred, on the basis of the clinical suspicion, to two genetic laboratories, the Department of Molecular Medicine and Medical Biotechnologies, University of Naples “Federico II”- CEINGE and the Pediatrics Unit, IRCCS Casa Sollievo della Sofferenza Hospital, S. Giovanni Rotondo (FG), to perform the genetic test. All these patients were referred before the beginning of this study. Among this 210 patients, 85 (40%) received a confirmatory diagnosis of MODY2 whether 125 (60%) were MODY2 negative proving a precision of the clinical suspicion of 40% (85/210) with 60% (125/210) of patients erroneously addressed to the genetic test (false positive at the clinical suspicion). To evaluate the clinical utility of the 7-iF in comparison to the selection made upon the clinical suspicion, we calculated the 7-iF score for all the patients of this “second retrospective cohort”. Sixty-four patients (64/210; 30%) were positive to the 7-iF (score: 7/7) and among them 52 were MODY2 proving a precision of the 7-iF in identifying MODY2 patients of about 81% (52/64) similar to the 76% of the prospective evaluation. Among the 7-iF negative patients, 77% (113/146) resulted non MODY2 and 22% (33/146) MODY2. These 33 patients, corresponding to the about 39% of all the MODY2 patients, would have been missed by the 7-iF. On the other hand, 113 patients out of the 125 patients addressed to the genetic test and resulted negative were correctly suspected by the 7-iF as non-MODY2 (about 90% of true negative) further confirming the clinical utility of the 7-iF in helping less experienced clinicians in suspecting MODY2 patients.

Discussion

The major issue with the MODY2 diagnosis is that, in general, physicians rarely suspect MODY2 and, also whether happen, many of them do not feel confident to ask for a molecular diagnosis. Some of the reasons can be the lacking of clear indications for the genetic test in the actual recommendations and the little experience that many physician have with rare monogenic disease. That might result in the estimated 80% of MODY2 patients without a correct diagnosed [9].

Here we propose a simple flowchart, the 7-iF, to identify candidates for the MODY2 genetic test. This flow-chart was conceived to be used by all physicians dealing with diabetes: it does not require specialized expertise in monogenic diabetes and is designed to be easily implemented in daily clinical practice in an outpatient setting. To explore its clinical utility, we evaluated its precision in prospective cohort and validated the results in two different retrospective cohorts. The 7-iF is broadly based on the EMGQN guidelines [5] but, unlike the latter, it includes T1D autoimmune markers and insulin treatment in the attempt to exclude patients with type 1 diabetes, which is the most prevalent form of diabetes (> 80%) in young individuals [11]. Consequently, a rationale search for different forms of diabetes is justified only when type 1 diabetes has been excluded. Although this strategy is not included in the EMGQN guidelines [5], many researchers and clinicians apply it [13], [14], [17], [18]. Regarding insulin treatment, most MODY2 patients have a modest impairment of glucose metabolism and usually do not need pharmacological intervention [17]. Therefore, patients with a positive history for insulin therapy are less likely to carry a pathological GCK mutation; consequently, only insulin-therapy naive patients are considered candidates for the genetic test.

The other additional criteria of the 7-iF with respect to the EMGQN guidelines are HbA1c levels, familiarity and age of onset. We used HbA1c instead of fasting glucose or the oral glucose tolerance test to select patients, because HbA1c reliably distinguishes between episodic hyperglycemia and persistent plasma glucose increase in the pathologic range. In fact, previous data [17], [19] and our clinical experience (Figure S1) indicate that a considerable proportion of MODY2 patients show a normal 2-hour glucose level at oral glucose tolerance tests. Indeed, recent guidelines for diabetes diagnosis include HbA1c as diagnostic marker [11], [20], although a consensus on the threshold value has yet to be reached. In the 7-iF, we use an HbA1c cut-off of 42 mmol/mol (HbA1c = 6%), which corresponds to a sub-diabetes state [20]. Based on the 7-iF, only patients with at least one parent affected by any alteration of glucose metabolism (type 1 or type 2 diabetes; gestational diabetes; IFG with or without IGT) are considered candidates for genetic testing. This criterion was included because MODY2 is an autosomal dominant inherited disease, with rare de novo mutations described [17]. Familiarity is not mandatory in the EMGQN guidelines [5], however recent studies found that it is the most powerful factor in distinguishing between MODY and sporadic forms of diabetes [10], [13], [14]. Furthermore, because MODY2 is usually discovered during infancy or in early adulthood, a diagnosis before 25 years of age is a criterion for the test in the 7-iF. Although a diagnosis of diabetes before 6 months of life is highly suggestive of a genetic cause, in such cases other types of monogenic diabetes (due to mutations in KCNJ11, INS or ABCC8 genes) are more plausible and should be evaluated before MODY2 [21]. Finally, the 7-iF excludes from the genetic test, patients with complex, severe clinical conditions, such as syndromic diseases, cancers, organ failures. Although a GCK mutation can co-exist with other diseases, patients with such features should be first evaluated by specialists in genetic diabetes to determine whether their clinical conditions cause the altered glucose metabolism or a pancreatic deficiency really occurs.

In the present study, we showed that the precision of the 7-iF was sufficiently high (76%) to make the tool appropriate for routine application, especially if compared to precision of the ‘clinical suspect’ (40% estimated from the second retrospective cohort). Furthermore, also considering the stringent 7-iF criteria we diagnosed a number of patients not smaller than that expect in a random diabetic population. In fact, we recommended the test to 17 patients of the prospective cohort and MODY2 was diagnosed in 13 of them. This corresponds to a MODY2 frequency of 1.4% (13/921), which is consistent with the expected frequency (0.56%–1.56%) [9], [15].

Despite the relatively high number of potentially missed cases with the application of the 7-if in place of the clinical suspicion (about 33% and 39% from the first and the second retrospective cohort, respectively), considering the ability of the 7-iF in correctly identify 76% of MODY2 patients in a general pediatric diabetic population, together with the excellent ability in correctly identify those patients with low chances of being MODY2 (about 90% of true negative), we believe that a widely application of the 7-iF to all the diabetic patients will result in an overall increase of the MODY2 diagnosis. That because, the availability of clear indications will increase the number of requested tests and will result in a decrease of negative results (for the higher precision of the 7-iF compared to the clinical suspicion). On the other hand, lower score of positive items to the 7-iF (6/7) should not be considered exclusion criteria. These latter patients would need further observation and eventually to be referred to center with experience in monogenic diabetes, especially if they present clinical features suggestive of genetic forms of diabetes, such as the familiarity consistent with a dominant disease, similar clinical features among familiars, and no necessity of insulin therapy in the proband and other familiars even after many years of disease.

Recently, Shield et al. [10] proposed and validated a mathematical model that estimates the probability that a diabetic patient has of being affected by any of the three most common forms of MODY, and reported the probability that each patient had of being a carrier of a pathological mutation. At variance, the 7-iF serves to identify patients potentially affected specifically by MODY2 (the most frequent MODY subtype reported in Italy [15], [16]) Interestingly, when we tested our retrospective cohorts for the latter proposed model [10] we obtained less satisfactory specificity (10%). This observation supports the use of the 7-iF as a useful tool to improve the diagnosis of MODY2. Particularly, providing as output a “yes or no” indication to the molecular test, it is a very friendly to use during a standard clinical examination by general practitioners.

During the prospective evaluation, several patients refused to undergo further specialized diabetes and genetic investigations. These patients, negative for autoimmunity markers, affected by mild diabetes, without on-going therapy and with a good glycaemia control, probably felt that further clinical evaluations were not necessary. Considering the clinical characteristic of the typical MODY2 patient, we believe this group to be enriched for MODY2.

Detailed analysis of 7-iF output revealed that each item included was frequent in MODY2 patients. In fact, almost all 7 items elicited a “yes” answer in more than 90% of patients. Interestingly, we found that most (> 90%) of the MODY2 patients met at least 6 of the 7-IF criteria and that the “missing” criterion was not always the same (Figure 2A). This result suggests that a positive response to any 6 of the 7 criteria of the 7-iF should be considered an indication for the genetic test, irrespectively of which item is missing. However, the precision of 6/7 items should be prospectively evaluated before being implemented in clinical practice.

The single item with the lowest prediction value was HbA1c level with sensitivity around 80%; this is probably due to the high threshold that we imposed. We selected a threshold that we believed would balance two opposite goals: to test all patients with a high suspicion of being MODY and to reduce the number of tests to those with highest probability of being positive. This threshold was chosen according to the International Expert Committee's definition of sub-diabetes status [20]. Of note, a relevant number (15%) of GCK-mutation carriers evaluated in the first retrospective cohort showed maximum HbA1c values slightly below 42 mmol/mol (i.e.<6%). while less than 1% had HbA1c levels above 64 mmol/mol (i.e.>8%) (Figure S2). Potentially, lowering the HbA1c threshold to 38 mmol/mol (HbA1c = 5.6%), or even using a range of HbA1c values from 38 mmol/mol to 64 mmol/mol (5.6%>HbA1c<8%), would improve the ability to correctly identify potential MODY2 patients to address to the GCK genetic test. However, these changes will need a prospective and retrospective validation in different cohorts and we expect a certain reduction of the specificity and of the positive prediction value.

We believe that the 7-iF is a useful tool for all physicians dealing with diabetic patients including those working outside the context of a specialized center. Although the 7-iF “yes/no”-approach forces a continuous clinical spectrum into artificial stringent threshold categories, it is probably the best way to make it standardized and replicable. The power of this method is that it does not require specific knowledge about this form of diabetes. In fact, it could induce non-specialist clinicians, who are less prone to consider MODY2 diabetes, to re-evaluate their strategy of diagnosis. An implicit caveat is that more complex and uncertain cases should be referred to specialized centers.

From a cost-effectiveness viewpoint, we believe that widespread application of the 7-iF would result in optimization of healthcare resources. By using QALY analysis (Material S1), we estimated that the application of the 7-iF could provide an increase in the quality of life and a saving for the health system of about 3000 in a 10 years period for every MODY2 diagnosed. Considering an expected prevalence of at least 1% of MODY2 diabetic patients among the Italian diabetic population [22] this would account for a 9 million euro saving per year. Therefore, the initial cost of the genetic test would be easily offset by the savings resulting from the avoidance of life-long insulin therapy and three-month follow-ups. The 7-iF criteria are highly specific for MODY2 and with no extra costs except the autoantibody evaluations for each patient, which are in any event recommended.

In conclusion, we provide the first simple set of binary items for the identification of MODY2 patients that has been retrospectively and prospectively validated. We demonstrate that the 7-iF is reliable and with high precision in identifying MODY2 patients. It can be easily implemented in all clinical settings to select patients to undergo the MODY2 genetic test. This strategy will probably increase the number of diagnoses of MODY2 with a consequent considerable impact on the patient's quality of life and on healthcare resources.

Supporting Information

Acknowledgments

This work did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector.

We would like to acknowledge all the patients and their referring clinicians. We thank Jean Ann Gilder (Scientific Communication srl., Naples, Italy) for writing assistance.

The following members of Italian Study Group on Diabetes of the ISPED contributed to the recruitment, to the clinical and the genetic counseling and to the molecular diagnosis for the referred patients included in the retrospective study. They are considered coauthors:

Martina Biagioni, Ines Carloni, Anna Maria Cester, Valentino Cherubini, Chiara Giorgetti, Antonio Iannilli: Division of Pediatric Diabetes, Dept. of Pediatrics, Salesi Hospital, Ancona, Italy.

Mariella Bruzzese, Francesco Mammì: Division of Pediatric Diabetes. Locri Hospital (RC), Italy.

Monica Guasti, Lorenzo Lenzi, Rossella Pepe, Barbara Piccini: Pediatric Diabetologic Unit, Meyer Children Hospital, Florence, Italy.

Marzia Benelli: Pediatrics Unit, Vito Fazzi Hospital, Lecce, Italy.

Francesco Cadario: Interdisciplinary Research Center of Autoimmune Diseases and Dept. of Medical Sciences, A. Avogadro University of Eastern Piedmont, Novara, Italy.

Valeria Calcaterra: Dept. of Pediatric Sciences, University of Pavia, Foundation IRCCS Policlinico San Matteo, Pavia, Italy.

Franco Cerutti, Sabrina Sicignano: Dept. of Pediatrics, University of Turin, Turin, Italy.

Corrado Mammì: U.O. Medical Genetics, A.O. Bianchi-Melacrino-Morelli, Reggio Calabria, Italy.

Nicola Lazzaro, Pasquale Comberiati: Pediatric Unit, San Giovanni di Dio Hospital, Crotone, Italy.

Andrea Scaramuzza, Gianvincenzo Zuccotti, Francesca Redaelli: Dept. of Pediatrics, University of Milan, Ospedale Luigi Sacco, Milan, Italy.

Francesco Gallo: Pediatrics Unit, “Antonio Perrino” Hospital, Brindisi Italy.

Marco Cappa: Patrizia Patera, Riccardo Schiaffini: Division of Endocrinology and Diabetology, University-Hospital Pediatric Dept., Bambino Gesù Children's Hospital, IRCCS, Rome, Italy.

Francesca Cardella, Caterina Salvo: Pediatric Diabetes Unit, Children's Hospital “G. Di Cristina”, Palermo, Italy.

Rosaria De Marco: U.O. Pediatrics, Hospital of Cosenza, Cosenza, Italy.

Margherita Chessa, Paola Frongia, Maria Rossella Ricciardi: Pediatric Diabetes Unit, Hospital Brotzu, Cagliari, Italy.

Carlo Ripoli, Maria Antonietta Zedda: Pediatric Unit, “U.O. Diabetologia dell'Età Evolutiva”, Cagliari, Italy.

Felice Citriniti: Pediatric Diabetes Unit, A.O. Ciaccio, Catanzaro, Italy.

Francesco Chiarelli, Stefano Tumini: Dept. of Pediatrics, University of Cheti, Chieti, Italy.

Maria Susanna Coccioli: Endocrinology and Pediatric Diabetes Unit, P.O. “D. Camberlingo”, Francavilla Fontana (Brindisi), Italy.

Fiorella De Berardinis, Elisa Santoro: U.O. Pediatric, A.S.P. Cosenza, Hospital of Cetraro-Paola, Italy.

Filippo De Luca, Fortunato Lombardo, Giuseppina Salzano: Dept. of Pediatrics, University of Messina, Messina, Italy.

Barbara Felappi, Elena Prandi: Dept. of Pediatrics, University of Brescia, Italy.

Elda Frezza, Elvira Piccinno, Cataldo Torelli, Clara Zecchino Endocrinology and Pediatric Diabetes Unit, Children's Hospital Giovanni XXIII, Bari, Italy.

Alfonso Galderisi, Carlamaria Monciotti: Dept. of Pediatrics, University of Padova, Italy.

Dario Ingletto: U.O. Pediatric, Hospital “Cardinale G. Panico”, Tricase (LE), Italy.

Peter Kaufmann, Bruno Pasquino: Dept. of Pediatrics, Regional Hospital of Bolzano, Italy.

Riccardo Lera: Diabetologic Unit, A.S.O. Alessandria, Italy.

Sonia Lucchesi: Pediatric Diabetes Unit, AUSL 6 Livorno, Italy.

Angelo Perrotta: Dept. of Pediatrics, A. Cardarelli Hospital, Campobasso, Italy.

Silvana Salardi, Mirella Scipioni: Dept. of Pediatrics, S. Orsola-Malpighi Hospital, University of Bologna, Bologna, Italy.

Sergio Luceri, Filomena Stamati: U.O. Pediatric – Castrovillari, Italy.

Luigi Pianese: Molecular Medicine Laboratory, Mazzoni Hospital, Ascoli Piceno, Italy.

Letizia Tomaselli: Institute of Internal Medicine, Endocrine and Metabolic Diseases, University of Catania, Italy.

Amedeo Vergerio (^†): Unit of Pediatrics Hospital of Feltre, Belluno, Italy.

Francesca Casaburo, Alessandra Cocca, Santino Confetto, Emilia Forgione, Ciretta Pelliccia, Stefania Picariello, Francesca Pisani, Alessia Piscopo, Pasquale Villano, Angela Zanfardino: Dept. of Pediatrics, Second University of Naples, Naples, Italy.

Pietro Buono, Adriana Franzese, Rosa Nugnes: Dept. of Pediatrics, University of Naples “Federico II”, Naples, Italy.

Giuliana Valerio: University “Parthenope”, Naples, Italy.

Claudio Maffeis, Marco Marigliano: Regional Center for Pediatric Diabetes, Clinical Nutrition and Obesity, Dept. of Life and Reproduction Science, University of Verona, Italy.

Giovanni Chiari, Brunella Iovene, Maurizio Vanelli: Regional Center for Children and Adolescents with Diabetes, Children Hospital, University Hospital, Parma, Italy.