2016, 63 (2), 187-191

Note SGLT2 inhibitors provide an effective therapeutic option for complicated with antibodies

Akinori Hayashi, Koji Takano, Sayuki Kawai and Masayoshi Shichiri

Department of Endocrinology, Diabetes and Metabolism, Kitasato University School of Medicine, Kanagawa 252-0374, Japan

Abstract. Diabetes mellitus complicated with insulin antibodies is rare in clinical practice but usually difficult to control. A high amount of insulin antibodies, especially with low affinity and high binding capacity, leads to unstable glycemic control characterized by hyperglycemia unresponsive to large volume of insulin and unanticipated . There are several treatment options, such as changing insulin preparation, immunosupression with glucocorticoids, and plasmapheresis, most of which are of limited efficacy. -glucose cotransporter 2 (SGLT2) inhibitors are a novel class of drug which decrease renal glucose reabsorption and lowers plasma glucose level independent of insulin action. We report here a case with diabetes complicated with insulin antibodies who was effectively controlled by an SGLT2 inhibitor. A 47-year-old man with treated with insulin had very poor glycemic control characterized by postprandial hyperglycemia unresponsive to insulin therapy and repetitive hypoglycemia due to insulin antibodies. Treatment with , an SGLT2 inhibitor, improved HbA1c from 8.4% to 6.0% and glycated albumin from 29.4% to17.9%. Continuous glucose monitoring revealed improvement of glycemic profile (average glucose level from 212 mg/dL to 99 mg/dL and glycemic standard deviation from 92 mg/dL to 14 mg/dL) with disappearance of hypoglycemic events. This treatment further ameliorated the characteristics of insulin antibodies and resulted in reduced insulin requirement. SGLT2 inhibitors may offer an effective treatment option for managing the poor glycemic control in diabetes complicated with insulin antibodies.

Key words: Insulin antibodies, Sodium-glucose cotransporter 2 inhibitors, Ipragliflozin, Continuous glucose monitoring

THE EMERGENCE of insulin-binding antibodies in administration [5, 6], plasmapheresis [6], use of anti- insulin-treated diabetic patients severely deteriorates CD20 antibody, and combination of the above have glycemic control characterized by severe hyperglyce- been advocated [5, 6], often with limited benefit. We mia unresponsive to high amounts of insulin injection report here a case with diabetes complicated with insu- and unanticipated hypoglycemic events. Insulin anti- lin antibodies that was effectively controlled by a bodies developed by repeated insulin administration sodium-glucose cotransporter 2 (SGLT2) inhibitor. bind to both exogenous and endogenous insulin and attenuates insulin action to elevate postprandial glu- Case Report cose level. Because the equilibrium between the bound and unbound insulin is not stable, large amounts of free A lean 47-year-old man was first diagnosed with insulin may be released abruptly from the insulin-anti- type 2 diabetes in 2003. His initial laboratory tests body complex (usually released from low affinity and revealed fasting/2-h glucose level 139/279 mg/dL and high binding-capacity antibodies) inducing unwanted serum C-peptide level 1.0/2.8 ng/mL, HbA1c 12.9 %, severe hypoglycemia [1-3]. Several treatment options and negative glutamic acid decarboxylase (GAD) anti- such as alteration of insulin preparations [4], steroid body. He had been treated with dietary therapy and oral hypoglyemic agents ( and ). Submitted Sep. 3, 2015; Accepted Oct. 12, 2015 as EJ15-0523 He was started on insulin therapy since 2007, initially Released online in J-STAGE as advance publication Nov. 7, 2015 with a pre-mixed insulin analogue, and then changed to Correspondence to: Akinori Hayashi, Department of Endocrinology, Diabetes and Metabolism, Kitasato University School of Medicine, multiple insulin injections (once-daily 1-15-1 Kitasato, Minami-ku, Sagamihara, Kanagawa 252-0374, and thrice-daily , total daily dose of 27 Japan. E-mail: [email protected] units) without experiencing any hypoglycemic episode ©The Japan Endocrine Society 188 Hayashi et al. until recently. In 2013, he developed anal canal carci- gliflozin, his CGM showed marked improvement with noma and was treated with TS-1 (tegafur, gimeracil, lower daytime glucose level and no nocturnal hypo- and oteracil potassium combination capsule) and mito- glycemia. His average glucose level was 151 mg/dL, mycin-C with concurrent radiotherapy. Shortly after glycemic SD 31 mg/dL, and the range in glucose level this treatment, he started to experience frequent epi- 81-221 mg/dL (CGM data not shown). Urinary glu- sodes of nocturnal hypoglycemia and daytime hyperg- cose was markedly increased from 3.4-4.0 lycemia. The addition of , discontinuation of g/day to 31.9-78.3 g/day, while urine volume was not insulin glargine, and use of a bedtime snack (160 kcal), significantly changed (from 1,450-2,400 mL/day to did not reduce hypoglycemic attacks nor control day- 1,280-1,650 mL/day). He stayed on once-daily oral time hyperglycemia. His HbA1c and glycated albumin 50 mg ipragliflozin and 50 mg sitagliptin, plus multi- (GA) was 7.9% and 26.0%, respectively. Continuous ple insulin injections (total daily dose of 15 units) with glucose monitoring (CGM) revealed nocturnal and no hypoglycemic attacks. CGM recorded 8 wk after predinner hypoglycemia with daytime hyperglycemia, with an average glucose level of 212 mg/dL, glyce- mic standard deviation (SD) of 92 mg/dL, and glucose range of 54-371 mg/dL (Fig. 1; black solid line). Other laboratory findings including blood cell count, serum albumin level, renal function, and liver function were all normal. Plasma concentrations of growth hormone, adrenocorticotrophic hormone, cortisol, thyroid stim- ulating hormone, free T3 and T4 level were all within normal range. GAD antibody, thyroglobulin antibody and thyroperoxidase antibody were negative. His serum C-peptide level was undetectable, while fast- Fig. 1 Glucose profiles before and after the treatment with ing immunoreactive insulin (IRI) level was 41.13 µU/ an SGLT2 inhibitor. Record of complete 48-h glucose mL and the insulin antibody titer was extremely high profile analyzed by CGM from the patient who (≥50.0 U/mL, 316 U/mL (assayed after 1/100 dilution); experienced frequent hypoglycemia and alternating binding ratio, 78.2%). Scatchard analyses of equilib- hyperglycemia due to insulin antibody (black solid line). rium-binding assay of insulin antibodies revealed cur- CGM recording was performed for 32 wks after addition of ipragliflozin (50 mg) to sitagliptin (50 mg) plus insulin vilinear response compatible with polyclonal antibod- therapy (gray solid line). Dotted lines represent glucose ies with a higher and lower affinity binding site, one levels of 70 and 140 mg/dL, respectively. with very low affinity and high binding capacity (K1 = 8 -1 -8 0.00105 × 10 M , b1 = 222 × 10 M), and the other with higher affinity and lower binding capacity (K2 = 8 -1 -8 0.0206 × 10 M , b2 = 80.4 × 10 M) (Fig. 2). These characteristics of his insulin antibodies rather resem- bled that of insulin autoantibodies found in patients with insulin autoimmune syndrome (IAS). Serological typing of human leukocyte antigen (HLA) alleles demonstrated the allelic combination of A2/24, B46/61, DR8/9. While a variety of insulin preparations such as , lispro, and were applied, his daytime hyperglycemia was not controlled and the frequency of hypoglycemic attacks was not decreased (HbA1c and GA were 8.4% and 29.4%, respectively). We finally added oral ipragliflozin (50 mg), an inhibitor Fig. 2 Reduction of binding affinity and binding capacity of the insulin antibody before (open circle) and after (closed of SGLT2, which controls blood glucose independently circle) the treatment with an SGLT2 inhibitor. Sera of insulin action [7, 8], while carefully adjusting insu- obtained from the patient were incubated with unlabeled lin doses by CGM. One wk after the addition of ipra- and 123I-labeled insulin and analyzed by Scatchard analysis. SGLT2 inhibitor for insulin antibody 189 the addition of ipragliflozin revealed that his average M). This treatment also reduced the insulin antibody glucose level improved to as low as 128 mg/dL with titer from 316 U/mL to 116 U/mL (both assayed after a glycemic SD of 38 mg/dL (CGM data not shown), 1/100 dilution). even though his fasting IRI level was 8.82 µU/mL and insulin antibody titer remained high (> 50.0 U/mL), Discussion with a binding ratio of 79.1%. Thirty-two wk after the addition of ipragliflozin, his insulin dosage was further We reported here a case where an SGLT2 inhibitor reduced to 8 units a day (insulin glulisine before each effectively improved glycemic control and ameliorated meal with no glargine) (Fig. 3). CGM showed addi- the characteristics of insulin antibodies in a patient tional improvement with average glucose level as low with diabetes complicated with insulin antibodies. as 99 mg/dL with a glycemic SD of 14 mg/dL (Fig. Presence of insulin antibodies induces high glyce- 1; gray solid line) without hypoglycemia. Glycemic mic variability. When there are high amounts of insulin control markers also improved with HbA1c from 8.4% antibodies with low affinity and high binding-capacity, to 6.0% and GA from 29.4% to 17.9%, respectively. they may induce severe hyperglycemia by annihilat- Analysis of his insulin antibodies performed 2 wk after ing the effect of endogenous and injected insulin, and the addition of ipragliflozin revealed marked improve- unexpectedly cause severe hypoglycemia by releasing ment (low affinity/high capacity sites: K1 = 0.00224 a large amount of free insulin [2]. The current treat- 8 -1 -8 × 10 M , b1 = 71.8 × 10 M; higher affinity/lower ment options including alteration of insulin prepara- 8 -1 -8 capacity sites: K2 = 0.0463 × 10 M , b2 = 16.8 × 10 tions [4], steroid administration [5, 6], plasmapheresis M) (Fig. 2). Continuation of ipragliflozin treatment for [6], use of anti-CD20 antibody, and combination of the 36 wk further reduced the number of insulin binding above [5, 6], are often of limited benefit. sites (low affinity/high capacity sites: K1 = 0.000905 In the present case, CGM showed glucose profile 8 -1 -8 × 10 M , b1 = 27.5 × 10 M; higher affinity/lower with high glycemic variability and Scatchard analy- 8 -1 -8 capacity sites: K2 = 0.107 × 10 M , b2 = 4.62 × 10 sis of insulin antibodies revealed similar characteris-

Fig. 3 Change of total daily insulin dose, HbA1c, and glycated albumin (GA) during the therapeutic course. 190 Hayashi et al. tics as observed in IAS [9] (Fig. 2). Drugs containing that control glucose fluctuations while reducing insu- the sulfhydryl group, such as methimazole, mercap- lin requirement. In the present case, alteration of the topropionyl glycine, and glutathion, and α-lipoic acid insulin preparation did not contribute to the reduction are known to be triggers of the onset of IAS. In the of insulin requirement, while the addition of an SGLT2 present case, the patient had no history of medication inhibitor successfully reduced the insulin requirement. with these drugs, and did not have HLA-DR4, which The reduction in insulin requirement may have sub- almost all Japanese patients with IAS possess [9]. sequently changed the characteristics and amount of Although several standard treatment options failed to insulin antibodies, leading to the successful elimina- control his glycemia, administration of ipragliflozin, tion of hypoglycemic attacks and insulin-insensitive an SGLT2 inhibitor [7, 8], which enhanced urinary hyperglycemia. glucose excretion, remarkably improved his glyce- The limitation of this study is that this is the first case mic control with less glucose variability and almost report that SGLT2 inhibitors provide an effective ther- completely eliminated his nocturnal hypoglycemia. apeutic option for diabetic patients complicated with Scatchard analysis revealed a curvilinear response sug- insulin antibodies. Further case reports must be accu- gesting polyclonal insulin antibodies with both high mulated to conclude the efficacy of this treatment option and low affinity sites (Fig. 2, open circle). The affinity because we cannot rule out the possibility of spontane- and the binding-capacity of the high affinity site rather ous remission of the disease process. We must also state resembled those of the polyclonal insulin autoantibod- that the usual precautions for the use of SGLT2 inhibi- ies detected in patients with IAS rather those of insu- tors in patients with type 2 diabetes must be exercised lin antibodies usually detected in patients treated with and careful adjustment of insulin doses by using CGM exogenous insulin injections. We speculated that the and/or frequent glucose monitoring must be applied to insulin antibodies with low affinity yet high binding- avoid hyperglycemic crisis and hypoglycemia, espe- capacity annihilated the insulin action when bound, cially in patients with low insulin-secreting capacity. and induced severe hypoglycemia when free insu- lin was released from the insulin-antibody complex Conclusion by some triggers. The immune reaction to exogenous antigen is enhanced when the expression of antigen is This is the first reported case demonstrating a dis- increased. The abrupt release of free insulin from the tinct benefit of ipragliflozin, an SGLT2 inhibitor, in immune complex further stimulates immune response patients with diabetes complicated with insulin anti- and increases the amount of antibodies which causes bodies. This case suggests that SGLT2 inhibitors pro- a vicious cycle. To terminate the vicious cycle it is vide an effective treatment option for managing the necessary to reduce the amount of antigen that causes poor glycemic control in diabetic patients complicated the immune response. When we reduced the amount with insulin antibodies. of exogenous insulin by the SGLT2 inhibitor, inducing the immune response to gradually subside, this might Disclosure have restored the inherent immune regulatory mecha- nism to reduce insulin antibodies. SGLT2 inhibitors None of the authors have any potential conflicts of decreases plasma glucose level independently of insu- interest associated with this research. lin action and have been shown to be an effective drug

References 1. Van Haeften TW, Krom BA, Gerich JE (1987) exogenous insulin. Endocr Rev 28: 625-652. Prolonged fasting hypoglycemia due to insulin antibod- 3. Radermecker RP, Renard E, Scheen AJ (2009) ies in patient with non-insulin-dependent diabetes mel- Circulating insulin antibodies: influence of continuous litus: effect of insulin withdrawal on insulin-antibody- subcutaneous or intraperitoneal insulin infusion, and binding kinetics. Diabetes Care 10: 160-163. impact on glucose control. Diabetes Metab Res Rev 25: 2. Fineberg SE, Kawabata TT, Finco-Kent D, Fountaine 491-501. RJ, Finch GL, et al. (2007) Immunological responses to 4. Lahtela JT, Knip M, Paul R, Antonen J, Salmi J (1997) SGLT2 inhibitor for insulin antibody 191

Severe antibody-mediated human insulin resistance: pheresis and prednisolone. Am J Med Sci 329: 259-264. successful treatment with the lispro. A 7. Chao EC, Henry RR (2010) SGLT2 inhibition--a novel case report. Diabetes Care 20: 71-73. strategy for diabetes treatment. Nat Rev Drug Discov 9: 5. Kawanami D, Ito T, Watanabe Y, Kinoshita J, Sakamoto 551-559. M, et al. (2013) Successful control of a case of severe 8. Hasan FM, Alsahli M, Gerich JE (2014) SGLT2 inhibi- insulin allergy with . J Diabetes Investig 4: tors in the treatment of type 2 diabetes. Diabetes Res 94-96. Clin Pract 104: 297-322. 6. Koyama R, Nakanishi K, Kato M, Yamashita S, 9. Hirata Y, Uchigata Y (1994) Insulin autoimmune syn- Kuwahara H, et al. (2005) Hypoglycemia and hyper- drome in Japan. Diabetes Res Clin Pract 24 Suppl: glycemia due to insulin antibodies against therapeutic S153-157. human insulin: treatment with double filtration plasma-