ORIGINAL ARTICLE

Safety and Efficacy of Metformin for Therapy-induced in Children With Acute Lymphoblastic Leukemia

Bruce Bostrom, MD,* Priya Uppal, BA,* Julie Chu, MD,* Yoav Messinger, MD,* Laura Gandrud, MD,w and Robert McEvoy, MDw

secretion in response to hyperglycemia and possibly Background: Hyperglycemia during corticosteroid and aspar- reduced number of insulin receptors from asparaginase.2,3 aginase therapy for acute lymphoblastic leukemia is a significant Treatment with either single agent corticosteroids or sin- side effect that is usually treated with insulin. Metformin is an oral gle agent asparaginase causes significant hyperglycemia in antidiabetic biguanide that may cause metabolic acidosis and liver about 1% of patients and the combination synergistically enzyme abnormalities of possible concern in patients receiving . increases glucose intolerance. Risk factors for hyperglycemia included age over 10 years, Down syndrome, obesity, and a Procedure: We reviewed patients with acute lymphoblastic leuke- family history of mellitus. The frequency of hyper- mia treated with corticosteroids and asparaginase who received glycemia increases when there is >1 risk factor.1 Pui et al metformin for control of hyperglycemia. observed hyperglycemia during remission induction in 41 Results: Seventeen patients received metformin, including 4 who (9.7%) of 421 pediatric patients who received the combination received insulin before starting metformin therapy. Twelve were of prednisone and asparaginase for ALL.4 In 39 of the 41 treated during initial induction therapy and 5 during relapse rein- patients, hyperglycemia developed within a week of the first duction. Corticosteroids included in 11, prednisone dose of asparaginase. Koltin et al5 observed a 15.7% incidence in 5, and megesterol in 1. Fifteen received pegasparaginase. of medication-induced diabetes in patients with ALL. They Patients were treated with metformin for a median of 6 days (range, confirmed the previously described association with age >10 2 to 46 d). Metformin was started at a median glucose level of years and Down syndrome. They also noted an increased risk 286 mg/dL (range, 112 to 499 mg/mL). The glucose level was con- trolled with metformin alone in 12 patients without the need for with central nervous system involvement at diagnosis. In con- insulin. Four patients received insulin before or concomitantly with tradistinction to earlier reports, they did not find any associa- metformin. In 1 patient, metformin failed to control the glucose tion with body mass index. In multivariate analysis, age >10 level, and insulin was administered. No significant toxicity from years and central nervous system involvement were sig- metformin was seen. Two patients had an elevated anion gap and nificantly associated with medication-induced diabetes. creatinine level because of extreme hyperglycemia. One patient had Transient hyperglycemia during induction chemo- mild elevation in total bilirubin and 5 patients had mild elevation in therapy for ALL is even more frequent, affecting up to 20% serum alanine aminotransferase levels. There were no episodes of of patients, as reported by Lowas and colleagues.6,7 Overt hypoglycemia. hyperglycemia during ALL induction has been associated Conclusions: Metformin is safe and effective for therapy-induced with decreased relapse-free survival and overall survival.8 hyperglycemia. Initially, insulin may be required for significant Traditionally, insulin therapy had been used for hyperglycemia or metabolic abnormalities. We are unaware of any treatment of symptomatic hyperglycemia in patients with prior studies using metformin in this population. ALL. Insulin requires subcutaneous administration and Key Words: metformin, leukemia, hyperglycemia, corticosteroids, frequent monitoring of the blood glucose level to prevent pegasparaginase hypoglycemia. Subcutaneous injections and finger pricks for glucose monitoring required with insulin therapy may (J Pediatr Hematol Oncol 2013;35:504–508) add more discomfort and stress to the child and family. Metformin, an oral antidiabetic agent is not associated with hypoglycemia, a known concern with insulin. In edication-induced diabetes mellitus, defined as glu- addition, metformin has many interesting off-target effects Mcose intolerance, which leads to hyperglycemia, that may be beneficial to patients with malignancies. A occurs commonly in children treated for acute lympho- number of studies have shown a lower incidence of malig- blastic leukemia (ALL) with a combination of cortico- nancy in diabetic patients treated with prolonged metfor- 9 and asparaginase.1 The mechanism is felt to be due min. Diabetic breast cancer patients on metformin show to insulin resistance from plus reduced an increased pathologic complete response after 21 weeks of neoadjuvant chemotherapy than diabetic breast cancer patients not on metformin.10 This may be due to the inhibition of AMPK by metformin and its downstream Received for publication December 5, 2012; accepted May 1, 2013. 11 From the Departments of *Pediatric Hematology/Oncology; and targets such as mTOR or AKT. Targeting mTOR has wPediatric Endocrinology, Children’s Hospitals and Clinics of shown benefit in childhood ALL.12 In addition, metformin Minnesota, Minneapolis, MN. may protect patients from anthracycline cardiotoxicity.13,14 The authors declare no conflict of interest. The anthracyclines doxorubicin and daunorubicin are fre- Reprints: Bruce Bostrom, MD, 2525 Chicago Avenue S. CSC 175, Minneapolis, MN 55404 (e-mail: [email protected]). quently given in conjunction with corticosteroids and Copyright r 2013 by Lippincott Williams & Wilkins asparaginase. Thus, patients may be placed on metformin

504 | www.jpho-online.com J Pediatr Hematol Oncol  Volume 35, Number 7, October 2013 J Pediatr Hematol Oncol  Volume 35, Number 7, October 2013 Metformin for Hyperglycemia in Leukemia therapy for hyperglycemia at the time they are receiving Pediatric endocrinologists were consulted for the cardiotoxic anthracyclines. The role of oral antidiabetics in administration of metformin therapy and management of adults with -induced diabetes is felt to be limited hyperglycemia needing intervention with insulin. In some because of slow action and a potential for side effects.15 A situations, the consulting endocrinologist elected to start comprehensive literature review did not reveal any reports insulin either before or simultaneously with metformin. on the use of metformin or other oral antidiabetic agents in pediatric patients with cancer or for medication-induced Data Analysis diabetes in children or adults with leukemia or lymphoma. Data were entered into SPSS for tabulation and descriptive analyses. MATERIALS AND METHODS RESULTS Patients Patients with ALL who developed hyperglycemia from Clinical Characteristics and Leukemia Therapy corticosteroids with or without asparaginase and received Patient demographics including age, weight, sex, eth- metformin with or without insulin were identified from the nicity, and ALL phenotype are outlined in Table 1. The electronic medical record data warehouse and pharmacy median age of 11 years in the 12 first remission patients was database. significantly higher than that of pediatric ALL patients, Glucose levels were monitored periodically while which peaks between 2 and 5 years of age. The distribution patients were on corticosteroid courses lasting longer than 7 of sex, ethnicity, and ALL phenotype were typical for our days, which typically occurs during remission induction or patient population. Family history for diabetes in parents, relapse reinduction. At our institution, at the onset of siblings, and grandparents was available in 15 of the hyperglycemia, if the patient is on intravenous fluids, dex- patients, of which 9 (60%) were positive. Documentation of trose is removed and the glucose level is monitored at least diabetes type in the family member was only available in 3 once daily. No attempt was made to control the patient’s (type 1 in UPN 7; type 2 in UPN 12 and 14). The phase of diet, as this is impractical because of the appetite stim- treatment and corticosteroid and asparaginase use are also ulation by corticosteroids. If patients were not on intra- documented in Table 1. All patients were being treated for venous fluids or hyperglycemia persisted after removal of initial induction (n = 12) or relapse reinduction (n = 5). All dextrose, immediate release metformin either once or twice but one were receiving a chemotherapeutic corticosteroid daily, depending on the dose, was instituted. (dexamethasone, n = 11 or prednisone, n = 5). UPN 16 After obtaining Institutional Review Board approval, was receiving megesterol (Megace) for appetite stimulation, the following data was retrospectively collected for analysis: which occasionally may result in hyperglycemia, especially name; medical record number; date of birth; sex; race/eth- at high doses.16 Pegaspargase was used in 15 patients and nicity; leukemia diagnosis date; treatment protocol; steroid was the only asparaginase product used. Two patients did used; asparaginase used; date metformin and/or insulin not receive pegaspargase because of allergy and were not started and ended; metformin dose; and family history of able to receive erwinase because of lack of FDA approval at diabetes mellitus. Serum glucose levels were obtained dur- the time. ing metformin and/or insulin therapy. All laboratory tests performed during, immediately before, and immediately Hyperglycemia Treatment after metformin administration were reviewed to assess Table 2 details the metformin treatment and results. toxicity. No special laboratory monitoring outside of that The median starting dose of metformin per day was routinely done during ALL therapy was carried out. 1000 mg (range, 500 to 2000 mg) or 28 mg/kg/d (range, 14

TABLE 1. Clinical Characteristics and Leukemia Therapy ALL Family History UPN Age (y) Sex Ethnicity Type of Diabetes Treatment Phase Corticosteroids Asparaginase 1 15.3 Female African American B Unknown Induction Dexamethasone Pegaspargase 2 3.1 Female White B None Induction Dexamethasone Pegaspargase 3 13.1 Male White B Grandparent Induction Prednisone Pegaspargase 4 2.8 Male White B Other Induction Dexamethasone Pegaspargase 5 15.2 Female White B None Induction Prednisone Pegaspargase 6 10.5 Female Mixed B Other Relapse induction Dexamethasone None 7 11.7 Female White B Other Relapse induction Prednisone Pegaspargase 8 17.4 Female White B None Induction Dexamethasone Pegaspargase 9 12.5 Male White T Grandparent Induction Prednisone None 10 9.8 Male White B None Induction Dexamethasone Pegaspargase 11 11.4 Male Asian B Unknown Induction Prednisone Pegaspargase 12 5.7 Female White B Father Induction Dexamethasone Pegaspargase 13 10.7 Female White B Grandparent Induction Dexamethasone Pegaspargase 14 12.4 Female White T Other Relapse induction Dexamethasone Pegaspargase 15 7.3 Male White B None Induction Dexamethasone Pegaspargase 16 12.6 Female Asian B None Relapse induction Megestrol Pegaspargase 17 13.2 Male White B Other Induction Dexamethasone Pegaspargase

ALL indicates acute lymphoblastic leukemia. r 2013 Lippincott Williams & Wilkins www.jpho-online.com | 505 Bostrom et al J Pediatr Hematol Oncol  Volume 35, Number 7, October 2013

TABLE 2. Hyperglycemia Treatment No. Days With Glucose No. Days to Glucose Glucose Level Starting Initial Level >200 Level at Peak at No. Metformin Metformin No. Days mg/dL Metformin Glucose Insulin Days From Weight Dose on After Starting Start Level Start on Start of Therapy to Control UPN (Kg) (mg/d) Metformin Metformin Alone (mg/dL) (mg/dL) (mg/dL) Insulin Insulin Hyperglycemia* 1 60.7 100 11 0 155 155 Metformin 2 11.7 750 3 2 270 332 Metformin 3 48.9 100 27 0 142 176 Metformin 4 16.8 500 6 Insulin used 419 584 419 8 0 Insulin + metformin 5 50.7 1500 11 2 303 366 Metformin 6 35.6 1000 9 0 299 299 Metformin 7 30.6 1000 22 3 198 253 Metformin 8 60.6 1000 11 Insulin used 112 200 212 4 3 after Insulin-metformin 9 36 500 19 0 250 250 Metformin 10 38.1 2000 21 Insulin used 428 428 465 1 1 after Insulin-metformin 11 37.6 14,000 2 0 218 218 Metformin 12 14.6 500 3 3 344 393 Metformin 13 33 1500 3 0 168 171 Metformin 14 70 1500 2 0 470 470 Metformin 15 30.9 500 46 Insulin used 291 906 522 11 8 before Metformin-insulin 16 29 500 4 Insulin used 499 1092 499 2 0 Insulin + metformin 17 36.1 1000 4 0 291 291 Metformin

*Insulin + metformin = simultaneous use of both; insulin-metformin = insulin followed by metformin; metformin-insulin = metformin followed by insulin.

to 64 mg/kg/d). The median number of days of metformin therapy for hyperglycemia. This patient developed hyper- treatment was 6 (range, 2 to 46). Usually metformin was glycemia during relapse reinduction, and metformin alone continued until after the corticosteroid therapy was com- effectively controlled hyperglycemia without the need for pleted. There were no episodes of hypoglycemia. insulin. The glucose level was controlled with metformin alone in 12 patients without the need for insulin. Four patients (UPN 4, Observed Toxicities 8, 10, 16) received insulin before or concomitantly with met- As part of routine management during induction formin. In 1 patient (UPN 15), metformin failed to control the therapy for ALL, patient undergo frequent testing of blood glucose level, and insulin was administered. The reason for counts and chemical analyses including electrolytes and insulin use varied. Tumor lysis–associated electrolyte dis- liver function tests. Table 3 details the results of anion gap, turbance and renal dysfunction in UPN 8 was considered a creatinine, bilirubin, and alanine aminotransferase levels in contraindication to initial metformin; therefore, insulin was these patients. Anion gap was used a marker of lactic started and metformin was added after metabolic abnormalities acidosis. Two patients (UPN 15 and 16) had mild elevations resolved. In the other 3 patients (UPN 4, 10, 16), the serum of creatinine and anion gap while on metformin or shortly glucose level at insulin initiation was over 400 mg/dL and the after it was discontinued. At the time, both had extreme consulting endocrinologist recommended insulin be used ini- glucose level elevations that peaked at 906 and 1092 mg/dL, tially based on past practices. Three of the patients (UPN 8, 10, respectively, and were treated with insulin accordingly. It is 15) started on insulin were weaned off insulin to metformin likely that the metabolic abnormalities were related to alone. Only UPN 4 required insulin to be continued for glucose ketoacidosis and were not a consequence of metformin. control. The only patient to fail metformin therapy was UPN UPN 1 had a mild elevation of total bilirubin level and 5 15. This patient was started on 250 mg bid of metformin for patients (UPN 1, 3, 4, 7, 8) had mild elevations of alanine a glucose level of 338 mg/mL, which subsequently increased aminotransferase levels. These elevations were of low grade to 906 mg/mL after 6 days, during which metformin was by common toxicity criteria v3.017 and are frequently seen increased to 500 mg bid. Nine days from the start of metfor- at the time of ALL diagnosis or during induction chemo- min, the glucose level was 522 mg/mL; therefore, insulin was therapy.18,19 Thus, we could find no conclusive evidence of started and metformin dose was increased to 500 mg tid with toxicity related to metformin use. rapid resolution of hyperglycemia. The insulin was dis- continued and metformin was continued at 500 mg bid until DISCUSSION the patient had completed his end of induction steroid taper. In the 12 patients who did not receive insulin, 8 had excellent Concern for Using Metformin in Patients control with metformin alone with no blood glucose levels over With Leukemia 200 mg/dL from the day after metformin was started. When we started using metformin for hyperglycemia Interestingly, UPN 6, who was diagnosed before we in ALL patients, there was a concern about increased tox- initiated the use of metformin for hyperglycemia in ALL icity due to concomitant chemotherapy or comorbid con- patients at our hospital, required insulin during induction ditions. This was not observed in any patient. In fact, there

506 | www.jpho-online.com r 2013 Lippincott Williams & Wilkins J Pediatr Hematol Oncol  Volume 35, Number 7, October 2013 Metformin for Hyperglycemia in Leukemia

TABLE 3. Observed Toxicities Maximum Elevated Maximum Elevated Creatinine Level Elevated Maximum Anion Total Bilirubin Level Elevated Maximum ALT Creatinine (mg/dL) Anion Gap (mEq/L) Bilirubin (mg/dL) ALT Level (U/L) UPN Level (normal <1) Gap (normal <17) Level (normal <1) Level (normal <65) 1 No No Yes 2.1 Yes 148 2 No No No data No data 3 No No No Yes 89 4 No No No Yes 143 5 No No No data No data 6No No No No 7 No No No Yes 83 8 No No No Yes 80 9No No No No 10 No No No data No data 11 No No No data No data 12 No No No data No data 13 No No No data No data 14 No No No No 15 Yes 1.6 Yes 19 No No 16 Yes 1.4 Yes 20 No data No data 17 No No No data No

ALT indicates alanine aminotransferase.

is likely more risk of toxicity due to hypoglycemia with the option. We do not feel elevated hepatic transaminases, use of insulin. Roberson et al20 described 6 patients with which can be seen in ALL patients, are a contraindication ALL who developed diabetic ketoacidosis and were treated to use metformin. with insulin. Four developed hypoglycemia because of over treatment with insulin. According to Oyer,15 “since proof of the benefits of tight control in steroid diabetes does not CONCLUSIONS exist, safety and avoidance of hypoglycemia remain equally In our experience, metformin is safe and effective for important goals.” Thus, metformin would be preferable controlling hyperglycemia induced from corticosteroids and because of the decreased risk of hypoglycemia. asparaginase and may obviate or reduce the need for insulin Lalau in a recent comprehensive review of metformin- with its attendant difficulties of hypoglycemia, subcutaneous induced lactic acidosis concluded that it is a rare complication administration, and frequent blood sugar monitoring. with an excellent prognosis when the drug is stopped. He In addition, metformin in combination with chemotherapy states, “Given the established efficacy of metformin, it would may theoretically improve the worse prognosis seen in ALL be paradoxical to deny the majority of patients with long patients who experience hyperglycemia by potentially inhibiting mTOR, an active target in leukemia cells. As established diabetes access to metformin because of the high 25 number of contraindications to this drug, when the same cardiotoxicity is a significant concern in ALL survivors, pharmacological mechanisms that relate to lactic acidosis the additional benefit of possible reduction in this risk would probably result in protective benefits to these would also be a welcome benefit of metformin. patients.”21 In adults with diabetes, acidosis was rare and never directly attributed to metformin.22 In cases of metfor- min overdose, acidosis was seen in 1.6% and hypoglycemia in REFERENCES 2.8% of patients.23 In that study, children under the age of 12 1. Howard SC, Pui CJ. Endocrine complications in pediatric years experienced few adverse outcomes, suggesting that patients with acute lymphoblastic leukemia. Blood Rev. younger patients are at lower risk for metformin toxicity 2002;16:225–243. during an overdose, which also may be true during ther- 2. Carpentieri U, Balch MT. Hyperglycemia associated with the therapeutic use of L-asparaginase: possible role of insulin apeutic use. Recent reviews suggest that metformin does not receptors. J Pediatr. 1978;93:775–778. appear to cause or exacerbate liver injury, and hepatotoxicity 3. Gailani S, Nussbaum A, Onuma T, et al. Diabetes in patients 24 is not a contraindication for its use. treated with asparaginase. Clin Pharmacol Ther. 1971;12:487–490. As this was an exploratory study, we did not have a 4. Pui CH, Burghen GA, Bowman WP, et al. Risk factors for formal clinical pathway or strict criteria for what level of hyperglycemia in children with leukemia receiving SC glucose would warrant the use of metformin and what level L-asparaginase and prednisone. J Pediatr. 1981;99:46–50. would require insulin. However, based on its lack of tox- 5. Koltin D, Sung L, Naqvi A, et al. Medication induced diabetes icity, we would suggest instating metformin for patients during induction in pediatric acute lymphoblastic leukemia: Support Care with a blood glucose level 300 mg/dL without significant prevalence, risk factors and characteristics. r Cancer. 2012;20:2009–2015. comorbidities such as tumor lysis syndrome, ketonuria, 6. Lowas SR, Marks D, Malempati S. Prevalence of transient or significant dehydration. For patients with a blood glu- hyperglycemia during induction chemotherapy for pediatric cose level >300 mg/mL or other comorbidities, a short acute lymphoblastic leukemia. Pediatr Blood Cancer. 2009; course of insulin followed by metformin would be an 52:814–818. r 2013 Lippincott Williams & Wilkins www.jpho-online.com | 507 Bostrom et al J Pediatr Hematol Oncol  Volume 35, Number 7, October 2013

7. Spinola-Castro AM, Siviero-Miachon AA, Andreoni S, et al. 15. Oyer DS, Shah A, Bettenhausen S. How to manage steroid dia- Transient hyperglycemia during childhood acute lymphocytic betes in the patient with cancer. J Support Oncol. 2006;4:479–483. leukemia chemotherapy: an old event revisited. Clin Adv 16. Aisner J, Tchekmedyian NS, Moody M, et al. High-dose Hematol Oncol. 2009;7:465–472. for the treatment of advanced breast cancer: 8. Sonabend RY, Siripoom VM, Okcu F, et al. Hyperglycemia dose and toxicities. Semin Hematol. 1987;24(suppl 1):48–55. during induction therapy is associated with poorer survival in 17. Common Terminology Criteria for Adverse Events (CTCAE) children with acute lymphocytic leukemia. J Pediatr. 2009; v3.0. Available at: http://ctep.cancer.gov/protocolDevelopment/ 155:73–78. electronic_applications/docs/ctcaev3.pdf. 9. Ruiter R, Visser LE, van Herk-Sukel MP, et al. Lower risk of 18. Segal I, Rassekh SR, Bond MC, et al. Abnormal liver cancer in patients on metformin in comparison with those on transaminases and conjugated hyperbilirubinemia at presenta- sulfonylurea derivatives: results from a large population-based tion of acute lymphoblastic leukemia. Pediatr Blood Cancer. follow-up study. Diabetes Care. 2012;35:119–124. 2010;55:434–439. 10. Jiralerspong S, Palla SL, Giordano SH, et al. Metformin and 19. Farrow AC, Buchanan GR, Zwiener RJ, et al. Serum pathological complete response to neoadjuvant chemotherapy aminotransferase elevation during and following treatment in diabetic patients with breast cancer. J Clin Oncol. 2009;27: of childhood acute lymphoblastic leukemia. J Clin Oncol. 3297–3302. 1997;15:1560–1566. 11. Shi WY, Xiao D, Wang L, et al. Therapeutic metformin/ 20. Roberson JR, Raju S, Shelso J, et al. Diabetic ketoacidosis AMPK activation blocked lymphoma cell growth via inhib- during therapy for pediatric acute lymphoblastic leukemia. ition of mTOR pathway and induction of autophagy. Cell Pediatr Blood Cancer. 2008;50:1207–1212. Death Dis. 2012;3:e275. 21. Lalau JD. Lactic acidosis induced by metformin: 12. Teachey DT, Grupp SA, Brown VI. Mammalian target of incidence, management and prevention. Drug Saf. 2010;33: rapamycin inhibitors and their potential role in therapy in 727–740. leukaemia and other haematological malignancies. Br J 22. Aguilar C, Reza A, Garcia JE, et al. Biguanide related lactic Haematol. 2009;145:569–580. acidosis: incidence and risk factors. Arch Med Res. 1992;23: 13. Ashour AE, Sayed-Ahmed MM, Abd-Allah AR, et al. 19–24. Metformin rescues the myocardium from doxorubicin-induced 23. Spiller HA, Quadrani DA. Toxic effects from metformin energy starvation and mitochondrial damage in rats. Oxid Med exposure. Ann Pharmacother. 2004;38:776–780. Cell Longev. 2012;2012:434195. 24. Brackett CC. Clarifying metformin’s role and risks in liver 14. Asensio-Lo´pez MC, Sa´nchez-Ma´s J, Pascual-Figal DA, et al. dysfunction. J Am Pharm Assoc. 2010;50:407–410. Involvement of ferritin heavy chain in the preventive effect of 25. Alvarez JA, Scully RE, Miller TL, et al. Long-term effects of metformin against doxorubicin-induced cardiotoxicity. Free treatments for childhood cancers. Curr Opin Pediatr. 2007; Radic Biol Med. 2012;57:188–200. 19:23–31.

508 | www.jpho-online.com r 2013 Lippincott Williams & Wilkins