A Genetic Risk Score for Insulin Resistance Identify Patients with Chronic Myeloid Leukemia, Treated with Nilotinib, Developing Diabetes

BACKGROUD: Nilotinib, a potent 2^nd generation tyrosin kinase inhibitor, is efficacious in the treatment of chronic myelogenous leukemia (CML). Impaired glucose metabolism represents one of the most frequently observed adverse events and several clinical trials, reported a high diabetes incidence during treatment with this drug. The mechanism of this side effect is poorly understood, but recently has been hypothesized an increased postreceptorial insulin resistence. Moreover, "in vitro"results indicated that c-ABL is involved in the insulin receptor signaling pathway. A large number of genetic variants associated with Type 2 diabetes (T2D) are implicated in beta-cell function but the role of common genetic variants associated with insulin resistance in the etiology of T2D, remains poorly documented. Scott R. and al. (Diabetes 2014) recently identified 10 multiple single nucleotide polymorphisms (SNPs) associated with insulin resistance and created a genetic risk scores (uGRS) as complementary tool to for insulin resistance.

AIM of the study was to identify whether this uGRS could be a valid prognostic tool in identifying patients developing diabetes during Nilotinib therapy

PATIENTS AND METHODS:45 patients (19 males) were included in the study. Twenty-four were treated with Nilotinib in first-line and 21 as second line (10 for resistance, 8 intolerance and 3 for other reasons). None of the subjects studied had abnormal blood glucose or took any antidiabetic drug before Nilotinib treatment.

We genotyped all patients with the GRS created by Scott R. including those in, or near, the IRS1, GRB14, ARL15, PPARG, PEPD, ANKRD55/MAP3K1, PDGFC, LYPLAL1, RSPO3, and FAM13A1 genes that have primary effects on subcutaneous adipocyte function. Also we added 2 new variants, one for TCF7L2 gene, associated with insulin secretion and another in FTO gene, whose effect on insulin levels is mediated by BMI. The uGRS was calculated, as previously reported, by summing the number of risk alleles across the 12 variants (0 for risk allele absent, 1 for heterozygosity and 2 for homozygosity for risk allele). A cut-off point for uGRS, maximizing predictivity and sensitivity of the score was calculated using Youden's index. Diabetes and impaired fasting Glycaemia were defined using the American Diabetes association (ADA) criteria. Data were reported as median and range,

RESULTS:Age at diagnosis was 45(18-82) years, the Sokal was low in 16 (42%), intermediate in 12 (26%) and high in 17 (32%) patients. During treatment and subsequent follow up of 45(7-127) months, 28 patients remained euglycemic, 5 (2 treated in the first line) developed diabetes after 14(1-32) months and 12 (6 treated as first line) developed IFG in 6 (1-12) months. No IFG patients developed an overt diabetes in the subsequent follow-up of 60.3 (33-102) months.

We calculated a cut-off point of 12 for uGRS. When the 28 euglycemic patient were compared with 5 patients becoming diabetics after Nilotinib, uGRS showed a sensibility of 100% and a specificity of 46% in predicting diabetes. Consequently, the positive predictive valuewas 25% where the negative predictive values 100%. When the 28 euglycemic patients were compared with the 12 patients developing IFG after Nilotinib, the sensibility and specificity of uGRS was low: 50% and 46% respectively.

CONCLUSIONS: A Genetic risk score for insulin resistance showed high specificity and a strong negative predictive values when used to identify CML patients treated with Nilotinib developing an overt diabetes. Further studies in lager populations are needed to confirm this predictivity that can be used in clinical practice to tailor the best anti TKI therapy.