Diabetes, hypertension, and cardiovascular events in survivors of hematopoietic cell transplantation: a report from the bone marrow transplantation survivor study

Improvements in hematopoietic cell transplantation (HCT) methods and supportive-care strategies have resulted in an expanding population of long-term survivors. Several recent review articles have provided comprehensive summaries of established late effects in HCT survivors^1-4 ; however, there are emerging issues in HCT survivors for which there exists a paucity of information. One such health concern that is prevalent in the general population, and that is becoming an issue of concern in cancer and HCT survivors, is the metabolic syndrome: a constellation of central obesity, insulin resistance, glucose intolerance, dyslipidemia, and hypertension. Metabolic syndrome is associated with a substantially increased risk for type 2 diabetes mellitus and atherosclerotic cardiovascular disease (CVD).^1-15  Data from the third National Cholesterol Education Program Adult Treatment Panel III (ATP III)^5,6  show the age-adjusted prevalence of metabolic syndrome in the US population of adults to be 26.7%.⁷  Additionally, a recent analysis of National Health and Nutrition Examination Survey (NHANES) 1999-2000 data demonstrates an increasing prevalence of the metabolic syndrome among US adolescents, where 6.8% of teens were found to have features suggestive of this syndrome.⁸ 

There is evidence to suggest that long-term cancer survivors may be at high risk for premature development of characteristics associated with the metabolic syndrome.^9-11  Cancer survivors are more likely to be significantly overweight, have higher fasting plasma glucose and insulin levels, and significantly decreased serum high-density lipoprotein (HDL) cholesterol levels.⁹  Furthermore, hyperinsulinemia, impaired glucose tolerance, hypertriglyceridemia, low HDL cholesterol level, and abdominal obesity were more common among the HCT patients than among a non-HCT group of leukemia patients or healthy controls.¹²  Additionally, certain cancer treatments are known to be cardiotoxic, most notably radiation therapy and common chemotherapeutic agents such as anthracyclines and cyclophosphamide.^13-18  Thus, the combined effects of metabolic syndrome and direct damage to the myocardium and endothelium by cancer therapy represent a potentially devastating late effect of cancer treatment.¹¹  Existing literature provides very little information regarding long-term cardiovascular outcomes in cancer survivors; however, an analysis of data from the Childhood Cancer Survivor Study has shown that the standardized mortality ratio for cardiac-related deaths was 8.2 (95% confidence interval [CI], 6.4-10.4) among long-term survivors of childhood cancer.¹⁹ 

Furthermore, while hyperglycemia and impaired glucose tolerance are well-recognized acute complications of cancer treatment, such as asparaginase and glucocorticoids,^20-23  or when steroids are used for prophylaxis or treatment of acute and/or chronic graft-versus-host disease (GVHD), there are minimal data regarding the long-term risk of diabetes in these patients. One study of 748 survivors of pediatric HCT found a higher than expected prevalence of both type 1 and type 2 diabetes in these survivors at a median of 11 years of follow-up.²⁴ 

The purpose of this analysis was to (1) ascertain the prevalence of obesity and medical late effects including diabetes, hypertension, and cardiovascular events in a cohort of patients treated with HCT for cancer or hematologic disorders; (2) to compare the prevalence of these medical late effects with the prevalence in age- and sex-adjusted siblings; and (3) to evaluate the association between these medical late effects, demographic, and clinical factors.

This analysis was part of the Bone Marrow Transplant Survivor Study (BMT-SS), a collaborative cohort study established in 2000 between City of Hope Cancer Center and the University of Minnesota. Eligibility criteria included (1) HCT between January 1, 1974, and December 31, 1998, at City of Hope or University of Minnesota; and (2) survival of at least 2 years after hematopoietic cell transplantation (HCT). This analysis was restricted to 1089 HCT survivors who were alive and not taking immunosuppressant agents at the time of the baseline interview. A random sample of 383 siblings of HCT survivors were also recruited for comparison. The informed consent process, study protocols, and documents were approved by the Human Subjects Research Review Committees at the City of Hope Cancer Center and the University of Minnesota, in accordance with the Declaration of Helsinki. Each eligible participant, or his or her parent or legal guardian if younger than age 18 years at the time of interview, provided informed consent for the study.

Participants, or their parents if they were younger than age 18 at the time of interview, completed the BMT-SS questionnaire, a 255-item survey assessing medical late effects, current medical conditions, medication use, health status, health behaviors, pregnancy history, demographic characteristics, socioeconomic indicators, and insurance coverage. The BMT-SS questionnaire asked participants to report impairment of organ systems and requests that the participant indicate the year the impairment was first diagnosed by a doctor or other health-care provider. This instrument was initially developed for use by the Childhood Cancer Survivor's Study²⁵  and has been subsequently modified to address topics related specifically to the HCT population in the current study. The BMT-SS questionnaire was validated in a random sample of 100 HCT survivors. The agreement with medical records was excellent (Kappa > 0.8 agreement adjusted for chance) for musculoskeletal impairments, cardiovascular disease (including myocardial infarction and hypertension), pulmonary conditions, endocrine conditions (including diabetes), and GVHD, and moderate for second cancers, central nervous system disorders, and eye problems (Kappa 0.4-0.7).²⁶  Additionally, self-reported height and weight have also been shown to be accurate and do not contribute significantly to errors in assessing body mass index (BMI).^27,28 

Five self-reported medical late effects were considered in these analyses: diabetes, hypertension (HTN), arterial disease, myocardial infarction (MI), and stroke. Participants were classified as diabetic if they answered “yes” to a question that asked about current use of a diabetic medication or if they reported onset of diabetes controlled by diet, pills, or insulin 2 or more years after their transplantation. Participants were classified as hypertensive if they reported current use of a medication to treat HTN or if they reported the onset of high blood pressure 2 or more years after transplantation. The presence of arterial disease was based on positive responses to questions that asked if participants had been told by a doctor or other health-care professional that they had hardening of the arteries or arteriosclerosis, coronary artery disease, or angina pectoris (chest pains due to lack of oxygen to heart requiring medication such as nitroglycerin), or if they indicated a coronary artery bypass surgery or angioplasty 2 or more years after their transplantation. MI and stroke were counted as events if participants indicated that they had these events 2 or more years after transplantation. HTN, diabetes, or cardiovascular events reported prior to HCT or within the first 2 years after HCT were excluded because of their likely association with medication use (calcineurin inhibitors and/or steroids for GVHD) or other early transplantation-related complications. Siblings were frequency matched for this analysis.

Case status (HCT survivors vs siblings), conditioning regimen, growth-hormone insufficiency, chronic GVHD, and donor type were considered as risk factors of interest for these analyses. Information regarding therapeutic exposures (conditioning regimens, prophylaxis, and treatment of GVHD), donor type, and presence of chronic GVHD was obtained from the HCT databases at each institution. Growth-hormone insufficiency was classified as positive if the participant indicated that they had growth-hormone insufficiency or received growth-hormone injections 2 or more years after their transplantation date. Age at transplantation, age at interview, race/ethnicity, sex, and BMI (calculated from self-reported height and weight at the time of the interview) were included as possible modifiers or confounders in the case-sibling analyses. Age at transplantation, age at interview, race/ethnicity, sex, BMI, and stem-cell source were included as possible modifiers or confounders in the case-case analyses.

The prevalence of medical late effects was described by tabulating “yes” responses to specific questions presented in the BMT-SS questionnaire. Within-survivor analysis was conducted to assess the role of clinical factors and therapeutic exposures such as conditioning regimen, growth-hormone insufficiency, chronic GVHD, and donor type. The analyses related to GVHD were limited to participants who received an allogeneic transplant. All analyses excluded subjects who reported the use of cyclosporine, steroids, or tacrolimus for at least 30 days in the past 2 years to account for the association between the known side-effect profiles of these medications that may themselves lead to the development of HTN and diabetes.

Descriptive statistics including means, standard deviations, frequencies, percentages, and ranges were calculated for demographic variables for the sibling comparison group and for demographic and treatment variables for the participating and nonparticipating HCT survivors. Two-sample t tests (for continuous variables) and chi-square tests (for dichotomous variables) were used to compare the difference between cases and siblings and between cases who participated and those who did not.

Frequencies and percentages were calculated for medical late effects among siblings and HCT survivors as totals and by conditioning regimen, growth-hormone insufficiency, chronic GVHD, and donor type among cases. The prevalence of each medical late effect, after adjusting for age at interview, age at transplantation, sex, race/ethnicity, and BMI was compared between cases and siblings by calculating odds ratios (ORs) and 95% confidence intervals using generalized estimating equations (GEEs) with a binomial distribution and a logit link. GEE methods were used in all models to account for the possible correlation between survivors and siblings from the same family.²⁹  In an analysis limited to survivors only, the use of total body irradiation (TBI), GVHD status, and donor type were evaluated in relation to the outcome variables in unconditional logistic regression models. These models were adjusted for age at transplantation, age at interview, race/ethnicity, sex, and BMI. Stem-cell source, growth-hormone insufficiency, and institution were not identified as independent predictors of the outcomes, nor did they appreciably alter the estimates, so they were not included in the final models. Interaction terms for age at transplantation, age at interview, sex, and conditioning regimen were also evaluated. Confounding by age at transplantation, age at interview, and sex was examined by looking at the strength and the precision of the estimate in both full and reduced models. SAS version 9.1 was used for all analyses (SAS Institute, Cary, NC).

Of the 2091 eligible HCT survivors, 1276 (83.1% of those successfully contacted) became study participants. Of the remaining survivors, 556 (28.8%) refused participation, 236 (13.2%) were lost to follow-up, and 23 (1.1%) enrolled but did not complete the health questionnaire. An additional 187 participants were not included in these analyses, as they reported either current or recent (past 2 years) use of immunosuppressant drugs. Characteristics of the study participants are shown in Table 1. Participants were more likely to be female (44.9% vs 37.6%, P = .003) and older at study participation (mean age, 39.3 vs 34.8 years; P < .001) than nonparticipants. Participants were also older (mean age, 30.8 years; range, 2 months to 68.6 years) than nonparticipants (mean age, 24.9 years; range, 1 month to 68.2 years) when they received their transplant (P < .001) and, on average, had survived 16 months longer than nonparticipants since their HCT (P < .001). Participants were more likely to report their race/ethnicity as white (P = .001), to have received an autologous transplant (P = .006), and to have had peripheral blood as their stem-cell source (P = .01).

When compared with the sibling comparison group, HCT survivors were more likely to be male (55.1% vs 38.9%, P < .001) and of nonwhite race/ethnicity (19.3% vs 6.0%, P < .001) but comparable in age at study participation (HCT survivors: mean = 39.3 years, standard deviation [SD] = 15.4 years; siblings: mean = 38.6 years, SD = 16.5 years) and in age at transplantation or the sibling's age at the time their brother or sister received a transplant (cases: mean = 30.8 years, SD = 16.7; siblings: mean = 29.4 years, SD = 17.5).

Among the participating HCT survivors, acute leukemia was the most common diagnosis (32.8%), and bone marrow was the most prevalent (68.4%) stem-cell source. Fifty-four percent of the survivors received an allogeneic transplant, and 73.8% received TBI as part of their conditioning regimen. Among those who received an allogeneic transplant, 40.8% developed chronic GVHD. Overall, 4.5% of the HCT survivors reported growth-hormone insufficiency or received growth-hormone injections.

The prevalence of medical late effects associated with metabolic syndrome by patient characteristics is shown in Tables 2–3. Diabetes was reported by 7.6% of HCT survivors and by 3.1% of siblings. While the specific type of diabetes could not be ascertained from data available in this study, of survivors with diabetes, 39.8% were obese (BMI ≥ 30 kg/m²) compared with 75% of siblings with diabetes who were obese (Table 4). The majority of both survivors and siblings with diabetes were treated by diet or oral agents (66.3% and 58.3%, respectively), and 26.5% of survivors and 33.3% of siblings were receiving insulin. HTN was reported by 18.5% of HCT survivors and by 14.9% of siblings. The prevalence of arterial disease, MI, or stroke was lower than 2% in the study population. The prevalence of obesity was 16.3% among HCT survivors and 19.8% among siblings.

In models adjusted for age at interview, age at transplantation, BMI, race/ethnicity, and sex (Table 5), HCT survivors of allogeneic HCT were 3.65 times (95% CI, 1.82-7.32) more likely to report diabetes than siblings, but the risk was not increased in autologous HCT survivors. Other factors that increased the risk of developing diabetes included Hispanic race (OR, 1.99; 95% CI, 1.12-3.55), older age at interview (OR, 1.14; 95% CI, 1.08-1.19), and BMI (OR, 1.13; 95% CI, 1.09-1.18). Interestingly, for each year older at the time of HCT, the odds of having diabetes decreased by 8% (OR, 0.92; 95% CI, 0.88-0.96). The odds of HTN were also significantly elevated in allogeneic HCT survivors (OR, 2.06; 95% CI,1.39-3.04), in those survivors who were older at the time of interview (OR, 1.07; 95% CI, 1.04-1.11), and for those who had a higher BMI (OR, 1.07; 95% CI, 1.04-1.10). As with diabetes, autologous HCT survivors were no more likely to report HTN than were siblings, and survivors of either allogeneic or autologous HCT were not more likely than siblings to report arterial disease, MI, or stoke.

In this analysis, models were adjusted for transplant type, sex, race/ethnicity, age at transplantation, age at interview, BMI, TBI exposure, and history of chronic GVHD, and these data are shown in Table 6. Compared with autologous transplant recipients, allogeneic transplant recipients were not more likely to report diabetes but were 2.31 times (95% CI, 1.45-3.67) more likely to report HTN. Exposure to TBI was associated with a 3.42-fold (95% CI, 1.55-7.52) increased risk of reporting diabetes but did not increase the risk for developing any of the other outcomes. Nonwhite race/ethnicity was associated with an over 2-fold increased risk for diabetes although a reduced risk for hypertension.

Other factors that increased the risk of developing diabetes and hypertension included older age at interview (OR, 1.14; 95% CI, 1.08-1.20 for diabetes; and OR, 1.06; 95% CI, 1.02-1.10 for HTN) and higher BMI (OR, 1.14; 95% CI, 1.08-1.18 for diabetes; and OR, 1.05; 95% CI, 1.01-1.08 for HTN). As seen with the case-sibling analyses, for each year older at the time of HCT, the odds of having diabetes decreased by 8% (OR, 0.92; 95% CI, 0.88-0.96). A history of chronic GVHD was not found to be associated with an increased risk for diabetes or hypertension.

With the exception of older age being associated with a higher risk of arterial disease (OR, 1.26; 95% CI, 1.10-1.44), we did not find an increased risk of arterial disease, MI, or stroke in association with TBI exposure, chronic GVHD history, or type of HCT. Participants who reported a history of growth-hormone insufficiency were 7.5 times (95% CI, 1.1-52.9) more likely to report stroke than those who did not report a history of growth-hormone insufficiency. However, the risk for other late effects did not vary by the presence or absence of growth-hormone insufficiency (data not shown).

This study provides long-term follow-up data on the largest cohort of HCT survivors ever examined for outcomes related to diabetes, HTN, and cardiovascular events. Our findings suggest that HCT survivors have an elevated risk of diabetes and HTN but that this population is not at an increased risk for stroke, MI, or arterial disease when compared with the sibling control group.

In this cohort, diabetes was more than 3 times as likely to occur in allogeneic transplant recipients as in a sibling comparison group, and when examined further, this risk was associated with exposure to TBI. The higher risk of diabetes in the allogeneic transplantation group, compared with autologous recipients, was also influenced by the fact that a higher proportion of those patients received TBI in their conditioning regimens. Previous studies indicate that radiation exposure to the abdomen may be associated with the development of diabetes in a small number of patients, but overall the evidence is quite limited.^30-32  Another study that compared 34 children and adolescents after either autologous or allogeneic HCT to 21 age-/sex-matched controls found that the 18 patients who received TBI had a significantly higher first-phase insulin response and insulinemia-glycemia ratio on glucose-tolerance testing compared with patients who received only lymphoid radiation, no radiation, or controls.³³  These findings suggest that TBI may play a role in the development of insulin resistance, which may predispose patients to the subsequent development of diabetes. Insulin resistance is also linked in the causal pathway leading to the cardiovascular effects of the metabolic syndrome because of its clinical association with increased rates of macrovascular disease and subsequent cardiovascular morbidity and mortality.^34,35  The only other large study in HC transplant recipients also found the prevalence of diabetes (6%, 35/578) to be higher after HCT than in the general population, and this prevalence was very similar to what was found in our study (7.6%, 83/1089).²⁴ 

HTN was not found to be influenced by TBI, but a significantly higher risk of HTN was associated with allogeneic transplantation. Interestingly, the risk of HTN was not increased in subjects who had a history of chronic GVHD where exposure to calcineurin inhibitors and steroids was likely more prolonged. Hypertension is a well-known side effect of calcineurin inhibitors while the exposure is ongoing, but whether this translates into a higher risk years after discontinuation of these medications is unknown but does not appear to be so based on our data.

Other mechanisms related to the immunologic and inflammatory effects of allogeneic HCT may play a role in the development of diabetes (or insulin resistance) and hypertension. Inflammation has been shown to be an integral component of the atherosclerotic process.³⁶  Elevated levels of C-reactive protein (CRP) predict coronary events in population studies and in individuals with angina,^37-39  and elevated levels of TNF-α and of IL-6 are correlated with worse outcomes.^40-42  Recent studies have shown a significant increase in inflammatory markers in association with both atherosclerotic disease and the insulin-resistance syndrome.^43,44  Inflammatory- and cytokine-mediated mechanisms that contribute to insulin resistance and development of features of the metabolic syndrome are of particular interest in individuals who have undergone an allogeneic HCT because the immunologic effects of these proinflammatory cytokines are mediators of the development and progression of both acute and chronic GVHD.^45-48  For individuals with chronic GVHD, this cytokine-mediated inflammatory process may continue for months or even years, and therapeutic maneuvers using cytokine inhibitory or blocking agents have recently been used successfully in treating chronic GVHD.^49-52  Interestingly, however, in our data the increased risk of diabetes and hypertension was not found to be directly associated with a history of chronic GVHD but was associated with receiving an allogeneic HC transplant regardless of whether the patients had a history of chronic GVHD or not.

The low prevalence of specific cardiovascular outcomes, such as arterial disease, MI, or stroke, can be explained by the fact that the study participants, both HCT survivors and siblings, are still relatively young with a relatively short mean follow-up of 8.3 years. Additionally, we did not identify risk factors for these events, with the exception of an increased risk of stroke in subjects who had growth-hormone insufficiency. Although details regarding pre-HCT therapeutic exposures, in particular, cranial radiation, are not available, it is possible that these subjects may have received additional cranial radiation that may have increased the risk of growth-hormone insufficiency as well as stroke.

Interestingly, HCT survivors were not found to be any more likely than the sibling comparison group to be obese, as might have been expected in view of their exposure to radiation, steroids, and prolonged periods of physical inactivity. A recent report from the Childhood Cancer Survivor Study (CCSS) examined obesity in 1765 adult survivors of childhood acute lymphoblastic leukemia compared with 2565 sibling controls. The age-/race-adjusted odds ratio for being obese in survivors treated with cranial radiation at doses of at least 20 Gy in comparison to the siblings was 2.59 for females (95% CI, 1.88-3.55; P < .001) and 1.86 for males (95% CI, 1.33-2.57; P < .001). Obesity was not associated with treatment consisting of chemotherapy only or with cranial radiation doses of 10 to 19 Gy.⁵³  Therefore, since the total dose of radiation exposure in HCT protocols is typically in the range of 10 Gy to 14 Gy,^54,55  it appears that HCT survivors who have not been exposed to additional cranial radiation might not acquire an additional risk of obesity. This is despite the fact that the dose rate is higher and the total treatment is given over 1 to 4 days, which may have a greater impact at the cellular level than a higher dose given over a longer time period.^56-58 

The results of the study must be interpreted in the context of potential limitations. Participation rate was 61% of those presumed eligible and 83.1% of those successfully contacted, and there were several differences between participants and nonparticipants as detailed previously. As a result, our analysis may have slightly overestimated or underestimated the prevalence of the outcomes of interest in the post-HCT period. We also acknowledge the fact that this study is based upon self-reported medical information and despite the fact that a validation study showed very good agreement between questionnaire answers and medical records, there is still the possibility that some subjects incorrectly reported conditions that they did not have or that they failed to report conditions that they did actually have. However, it is also important to note that the comparison group (ie, siblings) also provided self-reported data, thus there should not have been any systematic differences in bias by case or control status. Finally, there are other known risk factors for several outcomes in this study such as family history, smoking history, and physical activity that we did not have data available on and were not able to account for in the analysis that may have had an impact on the development of some of these complications.

Taken together, the findings presented here support the need for systematic and comprehensive research into the presence of early indicators of metabolic syndrome, or insulin resistance in HCT survivors, especially those who have received an allogeneic HC transplant.

Additionally, further studies that more precisely examine the mechanisms leading to the development of these outcomes will be important. HCT survivors should be monitored for the development of HTN and diabetes and counseled regarding health behaviors that might modify these outcomes, such as diet and exercise. Longer follow-up will be required to determine the true prevalence of serious cardiovascular events in this survivor population.

Contribution: K.S.B., as corresponding author, had full access to all the data in the study, had final responsibility for the decision to submit for publication and participated in the study design, analysis, interpretation of results, manuscript preparation, review, and submission; K.K.N. and S.B. participated in the study design, analysis, interpretation of results, and manuscript preparation, review, and submission; J.S., L.J.B., C.S., S.F., D.W., and J.G.G. participated in the interpretation of study results and manuscript preparation, review, and submission; A.C. and L.F. participated in the conduct of the study, data collection and preparation, and review of the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: K. Scott Baker, Department of Pediatrics, University of Minnesota, Minneapolis MN, 55455; e-mail: baker084@umn.edu.

This study was supported in part by grants from the National Cancer Institute (R01 CA078938; S.B.), the Leukemia Lymphoma Society (2192; S.B.), and the National Institutes of Health (K23 CA85503-01; K.S.B.).