Difficult-to-Diagnose Diabetes in a Patient Treated With Cyclophosphamide

The Contradictory Roles of Immunosuppressant Agents: A Case Report

Manuel García-Sáenz; Daniel Uribe-Cortés; Claudia Ramírez-Rentería; Aldo Ferreira-Hermosillo


J Med Case Reports. 2019;12(364) 

In This Article


Atypical diabetes accounts for up to 10% of cases of DM. A differential diagnosis is important in order to provide the best initial and long-term management, especially in patients with systemic comorbidities in which glucose elevations may jeopardize their quality of life and overall wellbeing. Initially, in this case, we had four diagnostic possibilities: drug-induced DM (due to prednisone that affects the metabolism of carbohydrates); ketosis-prone diabetes; type 1b DM, and finally a possible association with CY.

The use of glucocorticoids is widely associated with hyperglycemia. They stimulate the activity of phosphoenolpyruvate carboxykinase (PEPCK) in the liver and decrease its activity in the adipose tissue. This enzyme participates in liver gluconeogenesis and increases the level of glycerol 3-phosphate, increasing triacylglycerol (TAG) levels. In the adipose tissue, the inhibition of PEPCK decreases glyceroneogenesis, also decreasing TAG formation. Both, the increase in liver TAG and decrease in adipose tissue TAG formation induce serum elevation of free fatty acids, which are involved in the decrease of sensitivity to insulin in peripheral tissues with a consecutive rise in blood sugar.[10] However, at the time of evaluation, our patient was decreasing her prednisone dose. It should be noted that previously, a dose of high-dose intravenously administered steroids did not cause hyperglycemia. Furthermore, a glucocorticoid rarely precipitates DKA and patients with chronic treatment with a glucocorticoid also develop dermatological manifestations of insulin resistance such as acanthosis nigricans or acrochordons,[11] which were absent in this case.

Another entertained diagnosis was ketosis-prone diabetes. This type of diabetes was described in the 1960s in patients who maintain glycemic control without insulin therapy after an episode of DKA. These patients were obese and had a family history of diabetes, preserved insulin secretion, low prevalence of beta cell autoimmunity, and the ability to discontinue insulin therapy.[12] At the time of this publication (18 months after KDA), our patient seems to have recovered her β cell function completely, her weight is normal, she lacks clinical manifestations of insulin resistance, and she does not have a family history of diabetes, thus lacking the classical phenotype for this type of diabetes. In addition, the long-term recovery of endogenous insulin secretion (demonstrated with normal C-peptide levels) and the absence of pancreatic islet autoantibodies leads us to rule out T1DM.

After ruling out all common causes of diabetes, we considered that the hyperglycemic crisis may have been related to the CY treatment or precipitated by it. Previous reports of insulin-dependent DM in non-obese mice[13] also noted that CY, a cytostatic drug used for neoplastic and inflammatory diseases, may have a contradictory effect on immune regulation, affecting even the pancreas and insulin. At high doses it has been used for immunosuppression and at low doses it has been associated with enhanced immune responses through selective targeting of a subclass of suppressor T cells;[14] suppressor T cells are now referred to as Treg cells. Harada and Makino performed the first experiment that demonstrated that CY promotes the onset of DM in NOD mice.[15] Charlton et al. subsequently confirmed that CY accelerates T1DM development in NOD mice and that this phenomenon could be prevented through an increase in the amount of mononuclear cells.[16] The principal mechanism elicited by CY in this "acceleration model" was related to the depletion of a Treg cell population.[5–8] Specifically, it seems that CY induces a selective apoptotic loss of CD4+CD25+FoxP3+Treg cells from peripheral lymphoid tissues and the pancreas. Those Treg cells are involved in the suppression of the accumulation and function of effector T cells that attack pancreatic B cells.[5,7] Furthermore, CY could directly increase the number of interferon (IFN)-α producer T helper type 1 (Th1)-like cells capable of inducing islet destruction[17] and indirectly accelerate insulitis through enhancing IFN-γ secretion.[18] These mechanisms support the appearance of autoimmune diabetes, requiring a life-long use of insulin.

In our patient, traditional islet autoantibodies were negative, and she discontinued the use of insulin after 5 months. However, tests for other antibodies, such as anti-insulin receptor antibodies or Zn8, could not be performed. We suspect that CY could have precipitated hyperglycemia by a cytotoxicity mechanism, but other non-identified mechanisms may also have played a role. We also hypothesize that the effect may have been cytotoxic but not cytocidal, or that the surviving beta cells are currently enough to maintain a long-term normoglycemia and C-peptide production. The association of other factors predisposing to hyperglycemia cannot be completely ruled out in most humans, including our patient. After an exhaustive review of the literature, we only found one clinical case reported by Sharma et al.[19] of a 51-year-old woman with ductal invasive breast carcinoma who was treated with an epirubicin and CY (EC) regimen after surgery. After completion of four cycles of chemotherapy, she began to have clinical manifestations of diabetes, with a random glucose determination of 338 mg/dl and HbA1c of 12.4%. Interestingly, she also had no family history of diabetes and she improved both clinically and biochemically with glimepiride and metformin. As in our case, her history was not compatible with autoimmune diabetes. These authors also considered that there could be a possible relationship with CY use and the onset of DM.[19] We consider that the frequency of atypical behavior of diabetes associated with CY has been underdiagnosed for many reasons, one being that the severity of the disease leading to the use of CY is usually the focus of attention of physicians and diabetes becomes a lesser worry. Another reason may be that most cases present with milder forms of diabetes that do not trigger the alarm for atypical diabetes. Finally, the presence of other risk factors and drugs make it difficult to assign a specific value to the influence of CY in humans. More cases need to be published in order to increase the evidence of this influence.