In This Issue of Diabetes Care

A type of diabetes associated with the use of immune checkpoint inhibitor therapy appears to be a distinct form of diabetes, according to Wu et al. (p. 1292). While it is a rare complication, checkpoint inhibitor–associated autoimmune diabetes, or CIADM, appears to be characterized by relatively high rates of diabetic ketoacidosis at presentation and less frequent but earlier appearance of traditional type 1 diabetes autoantibodies. The findings come from a systematic review of individual cases reported in the literature with the aim of defining presentation characteristics, kinetics, and risk factors for CIADM. The authors then identified criteria for CIADM diagnosis as a combination of hyperglycemia and insulin deficiency, defined as C-peptide <0.4 nmol/L or diabetic ketoacidosis. A total of 192 individuals met the criteria for CIADM. Nearly all the patients identified had prior exposure to immune checkpoint inhibitor therapy for various cancer types. Of those tested, nearly 60% had susceptible haplotypes for type 1 diabetes, and median time to CIADM onset was 12 weeks. Of note, nearly 70% of individuals presented with diabetic ketoacidosis and just over 90% had low initial C-peptide, indicating significant insulin deficiency. Approximately 40% of individuals also had evidence of type 1 diabetes autoantibodies, and in these cases, there was a significant association between diabetic ketoacidosis and shorter time to CIADM onset. According to the authors, this may reflect the presence of a “primed” immune system that leads to a more severe response to checkpoint inhibitor cancer therapy. “It is not often that we are faced with an entirely new form of diabetes, and although we may be inclined to treat it exactly as we would type 1 diabetes, we are gradually learning about key differences that make this a flawed approach,” said author Linda Wu. “Our research synthesizes the granular details of reported cases of CIADM so that we can better understand and manage this emerging condition.”

Fetal growth in pregnant women with diabetes and obesity appears to be reduced in early pregnancy but is followed by overgrowth in later periods, according to van Poppel et al. (p. 1124). They propose that this biphasic growth pattern, and especially overgrowth in later pregnancy, is related to increasing prevalence of childhood overweight and obesity. Specifically, a buildup of triglycerides from early pregnancy and particularly in utero catch-up growth appears to increase the risk of later childhood obesity. The findings come from a narrative review of a series of large-scale longitudinal studies that looked at measurements of fetal growth in ∼14,400 pregnant women. They found that the growth pattern largely repeated across the studies in women with obesity, gestational diabetes, or type 1 diabetes. According to the authors, the patterns observed suggested that fetuses had undergone catchup growth in utero in much the same way as some infants grow post-birth. Consequently, they suggest such infants have a higher risk of obesity in later life. Exploring further, the authors found little evidence of variations in the patterns based on ethnicity or sex. They also considered possible delayed ovulation and measurement errors as potential modifying factors but ultimately conclude that the patterns are likely real. They next considered some of the potential reasons for early fetal growth reduction and examined the importance of this reduction. They propose that future studies should focus on these early growth patterns and raise the prospect of trying to improve lifestyles even at the prepregnancy stage in a bid to reduce large-for-gestational-age births and to achieve more consistent fetal growth prior to birth. Commenting further, author Mireille N.M. van Poppel said, “These results show that an adverse intrauterine environment affects the fetus already early in pregnancy, which has potential long-term consequences later in life. We hope to stimulate future studies that put emphasis on this important early phase in fetal development in order to better understand its role in childhood obesity.”

Early results from the Rare and Atypical Diabetes Network (RADIANT) study suggest that many new and rare forms of diabetes exist. According to the RADIANT Study Group (p. 1265), these atypical diabetes types likely include uncommon genetic syndromes as well as phenotypically distinct forms of diabetes that fall on a spectrum between the broad groups of type 1 and type 2 diabetes. According to early results, of 122 individuals with whole-genome sequencing completed, 3 individuals had a known diabetes monogenic gene and 6 had new monogenic variants. These variations were in the SMAD5, PTPMT1, INS, NFKB1, IGF1R, and PAX6 genes. There was also a series of phenotypic clusters related to lean type 2 diabetes, autoantibody-negative and insulin-deficient diabetes, lipodystrophic diabetes, and possible new forms of monogenic and oligogenic diabetes. In addition to reporting early results, the authors also describe the overall study design and motivations behind the study. The early analyses do not address the broader aims of the study (described in the article), which include wider application of genomic, phenotypic, and metabolic approaches as well as metabolic assessments. There are also many more individuals scheduled for inclusion in the study, with 878 individuals enrolled so far. According to Krischer, the analyses will lead to improved identification of atypical diabetes, further insights into the genomics involved, and potential specification of mechanisms/biomarkers via the metabolomics and transcriptomics they plan to include. They conclude that the study will likely have wide implications for defining and diagnosing such cases, with the knowledge generated potentially offering routes to prevent and treat the disease(s) in the future. Commenting further, author Jeffrey P. Krischer said, “RADIANT will also help providers recognize and distinguish among these different forms of diabetes, making possible more informed treatment decisions, improved patient outcomes, and, hopefully, a deeper understanding of the multiple pathways that result in this heterogeneous collection of diseases called diabetes.”

Adherence to diabetes medications appears to be greater in an American Indian population than previous estimates suggested, according to Scarton et al. (p. 1245). Adherence was also associated with improved glycemic control (HbA_1c), although the percentage of individuals at or below target remained below that of the wider U.S. population with type 2 diabetes. The findings come from an analysis of electronic health record data from the Choctaw Nation Health Services Authority and specifically an examination of HbA_1c levels and adherence to oral glucose-lowering medications. Just over 4,500 eligible individuals were included in the study, which represented approximately two-thirds of all individuals with type 2 diabetes in the health care system. They found that approximately one-third of the population was either below target (≤7%), above target (>7% to ≤9%), or uncontrolled (>9%) in terms of HbA_1c in 2017. Using proportion of days covered ≥0.80 as a proxy for medication adherence, they found that 66% of patients on biguanides (i.e., metformin) met the target. Equivalent estimates for other glucose-lowering drug classes included 72% for sulfonylureas, 75% for dipeptidyl peptidase 4 inhibitors, and 83% for sodium–glucose cotransporter 2 inhibitors. Most classes of drugs were associated with improvements in HbA_1c, although in one case (α-glucosidase inhibitors) the opposite was true. Notably, higher average adherence in 2017 corresponded with lower HbA_1c 1 year later. The authors report that the proportion of patients at or below HbA_1c targets increased from 32% in 2017 to 42% in 2018. They note that while their study removed cost as a barrier to adherence, they did not have discontinuation information or any hospital dispensing information. They also note that the study only covers 1 year. On that basis, they recommend that future studies be longitudinal and multiyear to estimate adherence more accurately—studies that the authors are reportedly considering.