Late Effects Screening of Acute Lymphoblastic Leukemia Survivors in the Military Healthcare System

ABSTRACT

Introduction

Pediatric acute lymphoblastic leukemia (ALL) survivors are a growing portion of the population with unique health screening needs. These survivors receive care within late effects oncology clinics and primary care clinics. Prior attempts to quantify compliance with follow-up recommendations have shown variable rates ranging from 28% to 73%. This study set out to assess rates of adherence to recommended health screening among pediatric ALL survivors within the U.S. DoD, identify potential risk factors contributing to patient compliance, and better define the prevalence of chronic health conditions.

Materials and Methods

This Institutional Review Board–approved, retrospective cohort study used data from the U.S. DoD MHS database and identified incident cases of pediatric ALL during 2007-2011 using a conservative case identification algorithm. Minimum duration of follow-up was instituted in order to ensure the entire study population had sufficient time for the assessment of each screening exam according to recommended guidelines. Rates of adherence to recommended screening measures were calculated across the full study follow-up period, and regression analyses assessed protective factors for compliance.

Results

One hundred and forty-four incident ALL cases were identified. During the follow-up period, 31.3% developed a new mental health diagnosis. In terms of recommended screening, 94.4% had an annual complete blood count for the entire study period, 90.3% had a liver function screening, 81.9% had an echocardiogram, 34% had a bone density scan, and 54.2% had a mental health visit. Adolescents were less likely to have a bone density scan (odds ratio [OR] 0.32, 95% CI, 0.11-0.95) or a mental health visit (OR 0.28, 95% CI, 0.11-0.7).

Conclusion

The MHS provides universal access to healthcare for all beneficiaries. In this population with universal access to care, there is increased compliance with screening recommendations. Our results reflect actual screening testing as opposed to general screening visits that have been previously reported in the literature. We also highlight the significant number of mental health diagnoses among pediatric ALL survivors.

Introduction

Acute lymphoblastic leukemia (ALL) is the most common cancer diagnosis in children. The treatment of pediatric ALL is a true success story of the last century. Until the mid-1900s, the disease was uniformly fatal, but with current treatment regimens, the overall survival rate is nearly 90%.¹ In 2011, a study estimated a prevalence of 35,982 survivors of pediatric ALL in the United States.² However, cure does come at a cost. According to cooperative group studies, a patient who received therapy for standard-risk ALL has a small but increased risk (1%) of developing a subsequent malignant neoplasm, a 70% likelihood of developing a chronic health condition by age 35 years, and a 16% chance of a severe or life-threatening chronic condition.³ Considering this, the Children’s Oncology Group (COG) Long-Term Follow-Up Guidelines were created to provide adequate screening recommendations for these conditions in survivors.⁴ In addition to an annual provider visit with mental health screening for all cancer survivors, specific testing is recommended based on treatment exposures. While screening recommendations are tailored to a patient’s disease and specific therapy regimen, given the near uniform use of antimetabolites, steroids, anthracyclines, and alkylating chemotherapy in the various regimens for the treatment of ALL, similar screening guidelines can apply to each of these survivors. Hepatotoxicity screening via one-time check of aspartate transaminase, alanine transaminase, and bilirubin is advised following exposure to antimetabolites and methotrexate. Screening for osteoporosis due to chronic steroid exposure is recommended using a one-time dual-energy X-ray absorptiometry (DEXA) bone mineral density exam. Per the COG guidelines, each of these exams is recommended within the first 2 years after completion of therapy. Screening for anthracycline- or radiation-induced cardiomyopathy is recommended within the first 2 years and then every 1-5 years depending on age when treated and cumulative doses of either therapy. Alkylator-induced hypogonadism screening by means of a one-time follicle-stimulating hormone/luteinizing hormone (FSH/LH) level during puberty is also advised. For the first 5 years following therapy, a regular schedule of screening complete blood count (CBC) is advised to monitor for disease recurrence.

Adherence to these recommended screening evaluations is dependent on several factors including provider knowledge of the recommendations and patient knowledge of their medical history, with specific knowledge of treatments, and willingness to pursue multiple consultation requests. To facilitate adherence, some patients receive care in dedicated survivor clinics where multiple consultations can be consolidated into the same visit. Other survivors may be followed in general oncology clinics or primary care settings, where easy access to laboratory and radiology testing or mental health screening may be less common. Survivors must often advocate for themselves in the primary care setting, as surveys of primary care providers show that they are often uncomfortable managing cancer survivors or unaware of recommended screenings for childhood cancer survivors.⁵

Prior studies investigating compliance rates with screening measures demonstrated follow-up rates with annual visits ranging from 27.8% to 91.2%, depending on the length of time from treatment.^6–11 Cardiovascular late effects of therapy can have a large impact on morbidity; one study demonstrated only 41.4% of patients received their recommended cardiac disease screening.¹² Previously identified factors associated with better adherence to guidelines include Caucasian race,^6,11 higher socioeconomic and insured status,^8,11 younger age at diagnosis,^6,9–11 proximity to care,^8,13 gender, history of relapse, brain tumor, sarcoma, or any concurrent late effects.¹⁴ Certain treatment factors, to include exposure to chemotherapeutic alkylating agents and radiation, have been associated with higher likelihood follow-up.⁸

The U.S. DoD MHS delivers care to 1.66 million active duty family members, 500,000 National Guard and Reserve family members, and 2.65 million retiree family members.¹⁵ An integral feature of the MSH/TRICARE is universal access to health care, delivered at little to no cost to DoD beneficiaries. These beneficiaries may receive their care either at civilian or military institutions across the United States and overseas. The purpose of this study was to evaluate the adherence with long-term follow-up screening recommendations and assess the prevalence of chronic health conditions among the pediatric ALL survivor population in the U.S. DoD. In addition, this study also sought to identify risk factors leading to poor compliance with screening among this population.

Methods

This was an Institutional Review Board–approved, retrospective, observational cohort study based on survivors of pediatric ALL in the DoD. The MHS Data Repository is comprised of healthcare records from treatment delivered in military treatment facilities as well as civilian facilities (via TRICARE billing) throughout the country for nearly 10 million beneficiaries. The database contains integrated administrative, medical, clinical, and pharmacy data on active and retired sponsors and their dependents.

The pediatric ALL survivor population in the study consisted of children and adolescents who were dependent beneficiaries between the ages of 2 and 19 years at diagnosis. Cases were identified through ICD-9 CM diagnosis codes for ALL during the 5-year period of January 1, 2007 to December 31, 2011. Cases then underwent a 2-year lookback in the system to ensure that a leukemia diagnosis had not been coded before the original diagnosis date to confirm these were new cases. A 5-year continuous enrollment in the database from diagnosis was required to ensure each cohort member could be followed for their 2-3 years of treatment followed by at least 2 years of post-treatment follow-up, thus allowing sufficient follow-up time for every patient in the study population to have received each screening exam of interest at least once. Infant ALL, which presents under the age of 2 years, has a different treatment and clinical course and was therefore excluded.

To ensure the identification of true ALL cases, each patient was validated by ensuring their record contained the expected procedure codes for bone marrow biopsy at diagnosis as well as intrathecal chemotherapy, and intravenous chemotherapy which occur throughout treatment. Subjects were excluded if they died during the study period or had diagnosis codes or evidence of subsequent chemotherapy consistent with either relapsed disease or a secondary malignancy. Subjects were also excluded if they had fewer than 2 years of chemotherapy, as this would suggest a deviation from a typical ALL treatment course. Given their unique treatment and outcome, a subset of patients with trisomy 21 was identified. Supplementary Table S1 presents the relevant codes used for study cohort identification.

Demographic and treatment-related variables were tallied to describe the study cohort. Age at the time of initial diagnosis was grouped by ages 2-10 years and 11-19 years for use as a proxy for current National Cancer Institute (NCI) risk categorization (standard versus high risk during this period). Sponsor rank was grouped as enlisted versus officer, serving as a proxy for socioeconomic status. Complete race and ethnicity data are unavailable in the database and thus not incorporated into the analysis. The following treatment-related variables were assessed through evidence of relevant procedure codes: radiation encounters, inpatient admissions, blood transfusions, and lumbar punctures. Radiation was categorized as yes/no while inpatient admissions, blood transfusions, and lumbar punctures were categorized as high/medium/low based on the respective interquartile range (IQR) derived from the tally of encounters. Late effects diagnoses that occurred during or after treatment were obtained via ICD-9 CM coded medical encounters before 2015 and ICD-10 CM coded medical encounters after 2015.

For individuals in the cohort, start of follow-up began 2.5 years after the diagnosis date for girls and 3.5 years after the diagnosis date for boys, due to the difference in duration of treatment based on standard of care during the study period. The following screening measures were assessed for compliance, based on the recommended frequency outlined in the COG Long-Term Follow-up Guidelines: annual visit with a CBC, one or more echocardiograms, one DEXA scan, one hepatic function panel, and one hypogonadism screening panel among patients 14 years or older (Supplementary Table S1 contains relevant procedure codes). Mental health screening was assessed by one or more visits to a mental health provider based on provider specialty taxonomy codes indicative of psychiatry, psychology, and social work.

For all descriptive data in the cohort, frequencies and percentages for categorical variables along with means, SD, medians, and IQR for continuous variables were tabulated. For risk factor assessment, regression analysis was performed for each screening measure. Bivariate analysis was conducted between patient characteristics and compliance rates for each screening measure using chi-square or Fisher’s exact tests, as appropriate. Age, sex, sponsor rank group, and type of care (military versus civilian) were included in all regression models, as well as variables that were significant at P < 0.1 in bivariate analysis. Stepwise logistic regression modeling techniques were employed to determine the final adjusted model. To assess for confounding, the final adjusted models were constructed retaining any variables, where removal resulted in a >10% change to point estimates. All data were analyzed using SAS, version 9.4 for Windows (SAS Institute Inc., Cary, NC).

Results

A total of 1,609 patients aged 2-19 years were identified based on an ALL diagnosis code occurring from January 1, 2007 to December 31, 2011. Of these, 686 (42.6%) patients were excluded as pre-existing, or prevalent, cases (via ALL diagnosis code present in the lookback period) or relapse (ALL relapse code during the study period). An additional 428 patients (26.6%) were excluded as they did not meet the 5-year continuous enrollment criteria that were necessary to include incident diagnosis, treatment course, and a minimum of 2 years from the end of treatment to assess follow-up screening. Of the remaining 495 patients, 144 (29.1%) were confirmed as incident ALL based on the confirmation scheme incorporating requirements for bone marrow biopsy, intrathecal chemotherapy, and ≥2 years of chemotherapy.

Clinical characteristics for the cohort are listed in Table I. The mean (SD) age of the cohort was 6.1 (4.2) years, with 80.6% (n = 116) of patients in the 2- to 10-year-old age group (i.e., NCI category of standard risk based on age). Females accounted for 43.8% (n = 63) of the cohort. Most patients (70.1%) were dependents of an enlisted sponsor, and the proportion of patients receiving care at a civilian facility versus MTF was 67.4% versus 32.6%, respectively. Six patients (4%) also had a diagnosis of trisomy 21. Treatment-related parameters aligned with the expected treatment course for pediatric ALL. The mean (SD) duration of chemotherapy treatment was 2.6 (0.5) years. The mean (SD) number of lumbar punctures was 19.4 (5.3). Patients receiving radiation accounted for 15.3% (n = 22) of the cohort, with mean (SD) of 7.5 (3.6) treatment encounters.

Incident late effects diagnoses that occurred while undergoing treatment or during follow-up are presented in Table II. Nearly one-third (31.3%) of the patients developed or were seen for one or more of the pre-specified mental health diagnoses: PTSD/adjustment disorder (18.1%, n = 26), anxiety (17.4%, n = 25), depression (6.3%, n = 9), and suicidal ideation (1.4%, n = 2). For endocrine late effects, 4.2% (n = 6) developed obesity, 0.7% (n = 1) developed osteoporosis, and 1.4% (n = 2) had delayed puberty. Prevalence of additional late effects diagnoses are as follows: 12.5% (n = 18) developed peripheral neuropathy, 6.3% (n = 9) developed either congestive heart failure and/or cardiomyopathy, 0.7% (n = 1) developed learning disabilities, and 6.9% (n = 10) were diagnosed with secondary acute myeloid leukemia; no patients had a diagnosis of cataracts.

The ALL cohort had a total of 895.8 years of follow-up time after the end of treatment; mean (SD) of 6.2 (1.7) years with a range of 2.2 to 9.8 years. The long-term follow-up screening measures within our cohort are described in Table III. This cohort demonstrated high compliance with CBC, hepatic function testing, and echocardiogram, with 94.4% averaging at least 1 CBC annually throughout follow-up, 90.3% having hepatic function testing, and 81.9% having an echocardiogram, with a mean (SD) number of 1.8 (1.1) echocardiogram encounters and an average (SD) time-to-first event of 2.2 (1.7) years. Compliance with DEXA scan was notably lower, with only 34% of patients completing this, with a mean (SD) time-to-first event of 3.0 (1.9) years. Only 23 (16%) patients were eligible for endocrinopathy screening (i.e., ≥14 years old), of whom 30.4% had an FSH/LH level measured. Lastly, 54.2% of patients had a visit with a psychologist, psychiatrist, or social worker during follow-up, with a mean (SD) number of visits and average time-to-first event of 10.3 (18.1) visits and 2.1 (2.0) years, respectively.

Bivariate analyses were independently performed for echocardiogram, DEXA scan, and mental health visit rates; these results are presented in Supplementary Table S2. These analyses were not performed for the annual visit with CBC and hepatic function screening due to near-complete compliance, 94.4% and 90.3%, respectively, and inadequate power given small sample size. Hypogonadism screening was not included in regression analyses due to the small number of patients eligible for that screening (n = 26).

The adjusted models per screening measure are presented in Table IV. For DEXA scan, the following variables were significantly associated (P-value <.05) with screening compliance: adolescent age at index diagnosis (i.e., 10- to 19-year-old age category), with an odds ratio (OR) and 95% CI of 0.32 (0.11-0.95) and anxiety diagnosis with an OR (95% CI) of 4.62 (1.68-12.66). For mental health visits, the following variables were significantly associated (P-value <.05) with screening compliance: adolescent age at index (i.e., 10- to 19-year-old age category), with an OR (95% CI) of 0.28 (0.11-0.70) and the composite measure of ≥1 of the assessed mental health diagnoses with an OR (95% CI) of 2.67 (1.19-5.99).

Discussion

Prior studies evaluating compliance with long-term follow-up metrics have used survey methodology and self-reported compliance or rates of follow-up visits in late effects clinics. Compliance rates with any screening guideline are multifactorial, including knowledge of recommendations and investment from both the patient and the clinician. In contrast, our study measured completion of recommended screening exams for late effects based on medical encounter claims from an integrated healthcare database. The MHS is a universally accessible healthcare system with little to no out-of-pocket expense for medical visits, which may enable patients to have improved compliance with follow-up measures.

In a 35-year follow-up report specifically for pediatric leukemia survivors from the Childhood Cancer Survivor Study, the reported prevalence of conditions directly caused by exposure to chemotherapeutic agents were as follows: secondary cancer 1%, cardiac conditions 1%, osteoporosis 1%, and cataracts 1%.³ Our study reflected a higher rate for several of these late effects diagnoses, with 6.9% having secondary acute myelogenous leukemia diagnosis and 6.3% developing a cardiac diagnosis. This may be related to a smaller sample size for our study. In addition, the increase in cardiac diagnoses, which can have a more indolent clinical presentation, may reflect a higher level of screening within the cohort. Our cohort did not have patients with cataracts, but this may reflect the shorter follow-up time. In their meta-analysis, Zhang et al. demonstrated the prevalence of obesity in ALL survivors to range from 34% to 46%, with one possible etiology being the long-term exposure to corticosteroids during treatment.¹⁶ Only 4.2% of our cohort developed obesity; however, obesity was identified via diagnosis codes and not through direct measurement of body mass index which would have likely reflected a higher rate.

Our study shows high adherence with annual CBC (94.4%), hepatic function panel (90.3%), and echocardiograms (81.9%). Rates for follow-up DEXA scan (34.0%) and screening for hypogonadism (30.4% of those eligible) were lower than we expected, although these rates have not been previously described. Our results demonstrate higher compliance overall than several previous studies that reported underutilization of follow-up services with rates ranging from 30% to 50%.^6–8 Our results for annual follow-up with a CBC and echocardiogram align more with studies that reported higher follow-up rates ranging between 70% and 80%.^9–11 Unique features of our cohort may explain this, such as the younger age and shorter follow-up time that overlapped with monitoring for disease recurrence. The universal access to care offered to military personnel and their dependents is an important factor that is likely to have had an effect on follow-up care, as is the access to multi-disciplinary survivorship clinics in the MHS which may have enhanced the rates of compliance.

Many prior studies assessed follow-up in oncology and late effects clinics, and these studies overlook the potential for primary care to take a role in the late effects monitoring of these patients. Our study did not place restrictions on where a screening procedure was performed; thus, our study captured not only patients in late effects oncology clinics but also patients who transitioned from an oncology department and continued their screening through their primary care team. One prior study described primary care providers’ sense of lack of knowledge about recommended screenings for cancer survivors; for these patients to be adequately cared for in primary care, there is a need for improved education on this subject in both graduate and continuing medical education.⁵

Although lacking in statistical significance, females and patients under 10 years of age at diagnosis seemed to have higher compliance with echocardiograms. Younger age at diagnosis, care in the direct care system, and officer rank were associated with higher compliance with DEXA scans. These results are consistent with prior studies showing that girls¹⁴ and survivors with a younger age at diagnosis^6,9–11 were more likely to be adherent to screening recommendations. In the MHS, rank can serve as a proxy for socioeconomic status,¹⁷ and this association was consistent with prior studies associating improved adherence with a higher socioeconomic status.^8,11

Nearly one-third of the cohort received care for a mental health diagnosis in the follow-up period, indicating the significant psychological toll that an ALL diagnosis and treatment can have. The Childhood Cancer Survivor Study demonstrated 17% of cancer survivors over the age of 20 years had a mental health diagnosis.² While there was a trend for increased mental health visits in males overall, there was a significant increase in mental health visits in individuals (male and female) who were diagnosed under the age of 10 years. This may reflect the universal access to care in the MHS providing greater access to mental health providers. It should also be noted that these rates and risk factors may differ from the general population due to different experiences for the military child. These experiences may include frequent relocations and family disruptions secondary to parental job assignments and/or deployment, which are additional established risk factors for mental health issues.¹⁸

Our study has several unique strengths. It was a large-scale database study that looked at both direct and civilian care for patients covered under the MHS. As a database study, it was not susceptible to survey bias where patients may over-represent their follow-up rates. An exhaustive look at the different components of follow-up care provided a multi-faceted view of the survivor experience, which has not been previously documented in a single study.

The limitations of our study are largely related to our sample size as well as the potential for misclassification bias. Of the 923 non-prevalent, non-relapse cases, only 144 (15.6%) were included in the final cohort, but each of the steps of attrition was deemed essential to identify true leukemia patients with observable follow-up time. Sample size limitations impacted the regression analysis power, and while the identified risk factors support what has been previously published, few held statistical significance. Comparison of care delivered in military versus civilian institutions was also limited due to sample size. Screening adherence rates for CBC, hepatic function testing, and hypogonadism screening were unable to be performed due to limited power from our small sample size. In relying solely on diagnosis and procedure codes, there was the risk of misclassification bias. Each of these tests identified in our cohort could have been obtained through routine screening of asymptomatic patients or for evaluation of symptomatic patients.

Conclusions

The universal access to care, including cancer survivor clinics and resources in the MHS, allowed for improved screening rates of the population at risk. Our results, although not significant, keep with a trend of older males being at risk for being lost to follow-up. These specific groups should be a target of future clinical studies for education and delivery of survivorship screening. The mental health burden in this population was also highlighted, as nearly one-third of the cohort developed mental health diagnoses. Younger patients and boys appeared to have an increased risk for developing a mental health diagnosis, reflecting that this population may be particularly at risk. Care of the cancer survivor demands a multi-disciplinary approach including pediatricians and primary care providers, oncologists, mental health providers, and relevant pediatric subspecialists.

Acknowledgment

None declared.

Supplementary Material

Supplementary Material is available at Military Medicine online.

Funding

None declared.

Conflict Of Interest Statement

None declared.

References

Author notes