Physician-diagnosed depression and suicides in pediatric hematopoietic cell transplant survivors with up to 40 years of follow-up

Although childhood hematopoietic cell transplantation (HCT) survivors often experience acute psychological effects,¹ very little is known about their long-term mental health. In a previous quality-of-life study of our long-term childhood HCT survivors, depression assessed on standardized self-reported measures was higher in patients compared with sibling controls² suggesting that depression may be an important psychological complication, but specific information is lacking. Among childhood cancer survivors, the prevalence of depression ranges from 5 to 30%^3–6 and depression has been associated with female gender,^5,6 cranial irradiation⁵ and obesity.⁴ In childhood HCT survivors, the prevalence of depression is largely unknown with one retrospective study reporting 12% of survivors having unspecified signs of depression⁷ and depression has been associated with prednisone,⁸ female gender² and the amount of therapy received before transplant.² Childhood and young adult cancer survivors are also more likely than controls to use antidepressants.^9,10 The purpose of this retrospective study was to determine the prevalence and identify risk factors for physician-diagnosed depression or suicide in childhood HCT patients.

Between 1969 and 1 January 2011, 1122 consecutive children < 18 years of age received conditioning therapy and survived at least 2 years after HCT at the Fred Hutchinson Cancer Research Center (FHCRC), Seattle, WA, USA. Excluded from the analysis were 18 patients who did not consent to follow-up or who were incarcerated, 14 with depression diagnosed before referral for transplantation, and 6 with metabolic disorders that could predispose them to depression. The remaining 1084 patients or their responsible guardians consented to follow-up and data were reviewed under a protocol approved by the FHCRC Institutional Review Board.

Patient and transplant characteristics are presented in Table 1. All patients returned to FHCRC for long-term follow-up at 1 year and electively thereafter. During transplantation and follow-up exams, depression was assessed by the attending oncologist. Following transplantation, all patients were followed by the FHCRC’s Long-term Follow-up Program. Follow-up consisted of annual phone contact with referring physicians from 1969 to 1990 and by questionnaires mailed annually to patients and their primary medical provider from 1991 to the present. Questionnaires asked about new diagnoses and medications. In addition, patients were selectively contacted to obtain more complete medical information and/or medical records. The annual response rate to follow-up by patients and/or their physicians was 30–50%. Overall, 80% of patients had been contacted within 3 years of the study date. Follow-up data were entered into a late effects database. Cases of depression were reviewed and defined as a physician diagnosis of depression. For the purpose of analysis, three patients with mild depression who did not require therapy were not considered as cases. In addition, four patients who committed suicide were included as cases as they had probable depression, but had not been diagnosed by a physician. Temporary iatrogenic depression caused by steroids or interferon was excluded. Data were insufficient to further categorize depression as major or minor depressive disorder. Obesity was defined as a body mass index (BMI) ≥ 95th percentile based on age and gender in patients < 18 years old and a BMI ≥ 30 in patients ≥ 18 years old. Surgery after transplant was defined as any major surgery and categorized as orthopedic or nonorthopedic. Cumulative incidence curves for the development of depression were estimated treating death without depression as a competing risk. Cox regression was used for hazard ratio analysis with adjustment for covariates. Variables found to be nonsignificant in univariate analysis included transplant diagnosis, transplant regimen, therapy before transplant, cranial irradiation, hepatitis C virus infection, diabetes, growth hormone deficiency, second cancer, thyroid hormone replacement and pregnancy. All other variables significant (P ≤ 0.05) in univariate analysis were entered into the multivariate model, except for gender (P = 0.07) which was included for its known association with depression. Because prednisone therapy and chronic GvHD were found to be highly correlated, only prednisone was selected to be included in the multivariate analysis as it was the more significant variable. Descriptive statistics were described as median and range.

Table 1.

Patient and transplant characteristics of study population

	Diagnosed with depression	Total
Number of patients	194	1,084
Age at transplant—years median (range)	11.8 (0.6–17.9)	9.0 (0.3–17.9)
Gender
Female	93 (48%)	453 (42%)
Male	101 (52%)	631 (58%)
Race/ethnicity
Non-Hispanic White	166 (86%)	856 (79%)
Other race/ethnicity	28 (14%)	228 (21%)
Diagnosis at transplant
Hematologic malignancy
Acute lymphoblastic leukemia	52 (27%)	332 (31%)
Acute myelogenous leukemia	47 (24%)	210 (19%)
Chronic myelogenous leukemia	20 (10%)	87 (8%)
JMML/MDS/MPD	10 (5%)	62 (6%)
Hodgkin Disease	6 (3%)	18 (2%)
Non-Hodgkin lymphoma	6 (3%)	25 (2%)
Nonhematologic malignancy
Neuroblastoma and other solid tumorsa	7 (4%)	108 (10%)
Nonmalignant disorder
Aplastic anemia	41 (21%)	159 (15%)
Other nonmalignanta	5 (3%)	83 (8%)
Donor type at first transplant
Autologous	18 (9%)	158 (15%)
Sibling/twin	107 (55%)	565 (52%)
Related (parent or relative)	29 (15%)	92 (8%)
Unrelated	40 (21%)	269 (25%)
HLA matching at first transplant
Matched	139 (72%)	750 (69%)
Mismatched	37 (19%)	176 (16%)
TBI at first HCT
High-dose 12–15.75 Gy	133 (69%)	667 (61%)
Low-dose 2–8 Gy	3 (1%)	52 (5%)
None
Busulfan–cyclophosphamide	21 (11%)	136 (13%)
Cyclophosphamide	30 (15%)	125 (12%)
Other chemotherapy	7 (4%)	104 (10%)
Acute GvHD grades
0–I	87 (45%)	396 (37%)
II–IV	89 (46%)	530 (49%)
Second transplant	19 (10%)	144 (13%)
Clinical extensive chronic GvHD (yes/evaluable)b	79/176 (41%)	331/926 (36%)
Survivors	143 (74%)	813 (75%)
Follow-up—years median (range)	23.3 (2.9–40.1)	16.5 (2.0–40.1)
Age at onset of depression—years median (range)	21.3 (7.3–52.7)	—
Years after HCT at onset of depression—years median (range)	12.0 (0.1–37.7)	—

Abbreviations: HCT = hematopoietic cell transplantation; JMML = juvenile myelomonocytic leukemia; MDS = myelodysplastic syndrome; MPD = myeloproliferative disorder.

^aNonhematologic malignancies included neuroblastoma (n = 81), brain tumors (n = 9), Ewing sarcoma (n = 8), Wilms tumor (n = 5), germ cell (n = 4) and retinoblastoma (n = 1). Other nonmalignant disorders included primary immunodeficiency (n = 42), red cell dysplasia (n = 24) and other miscellaneous diseases (n = 17).

^bExcluded were autologous only HCT recipients.

As shown in Table 1, depression was identified in 194 patients (190 by physician diagnosis and in 4 patients who committed suicide). Of the 190 patients with physician-diagnosed depression, 180 patients had unipolar depression and 10 had bipolar disorder. The cumulative incidence of depression was 39% at 40 years after HCT. Of the 194 patients with depression, 113 patients had a history of chronic GvHD for which 93 received prednisone therapy (26 patients were diagnosed with depression while receiving prednisone and 67 patients had completed prednisone therapy.) Before diagnosis of depression, 70 patients had one or more surgeries after transplant, including 36 orthopedic and 41 nonorthopedic related surgeries. Orthopedic surgeries included: core decompression or hip replacement for avascular necrosis (n = 14), knee surgery (n = 6) and other (n = 16). Nonorthopedic surgeries included: appendix (n = 8), cholecystectomy (n = 7), brain surgery (n = 7), hysterectomy (n = 3), organ transplant (n = 3), splenectomy (n = 3), cancer-related surgery (n = 3) and other (n = 7).

Four male patients committed suicide at 3.1, 13.5, 22.2 and 28.5 years after first HCT at 19, 22, 35 and 42 years of age. None of these patients had a known history of diagnosed depression prior to committing suicide. In addition, nine patients (five males and four females) who were diagnosed with depression at 19.9 (8.3–34.4) years of age attempted suicide 7.0 (0.8 to 26.5) years after HCT at 20.1 (16.1 to 35.2) years of age. The rate of suicide or suicide attempt was 1.2% (13/1084), which is consistent with a report of the incidence of suicide or suicide attempt within the first 6 years after adult HCT of < 1%.¹¹

Among the 190 patients with physician-diagnosed depression, 166 (87%) initially received antidepressants and/or psychotherapy, 13 received only psychotherapy and 11 were prescribed therapy but were not treated. At last follow-up, 85 survivors had received antidepressants (61 one drug, 21 two drugs and 3 three or more drugs) for 8.5 (0.2 to 23.7) years. Among the 190 patients, 47 patients died and depression resolved in 48, leaving 95 survivors with persistent depression. Overall prevalence of persistent depression was 12% (95/813) among current survivors, 15% (87/601) among adults and 4% (8/212) among survivors < 18 years old. The 12% overall prevalence of depression was within the 5–30% range reported among childhood cancer survivors.^3–6 The 15% prevalence of depression among adult survivors was twofold higher than the expected 7.1% in the general population,¹² and the 4% among child survivors was higher than the expected 0.5–3.5% in the general population.¹³

The results of multivariate analysis are presented in Table 2. Depression was related to several risk factors present at transplant and occurring months to decades after transplantation. The risk of depression was higher among older children, especially 12–17 year olds, which is consistent with a prior study that showed adolescent (≥ 13 years of age) childhood cancer survivors have higher levels of post-traumatic stress and worse mental health scores than younger patients.¹⁴ Depression risk was also higher among recipients of donor grafts from parents or relatives compared to sibling donors, and the majority of parent/relative donors were HLA mismatched, suggesting that depression may be related to the development or treatment of chronic GvHD. Depression has been previously associated with obesity in childhood cancer survivors⁴ and with an ~ 25% increase in lifetime diagnosis of major depression in adults in the US population.¹⁵ Depression has been associated with prednisone exposure among long-term allogeneic HCT survivors.⁸ Depression was associated with post-transplant surgery, including orthopedic surgery, frequently for avascular necrosis and nonorthopedic surgery, of which several indicated major medical problems. Growth hormone deficiency is a frequent complication of the use of TBI in childhood HCT survivors. In our study, depression was not associated with growth hormone deficiency per se, but was associated with the subset of patients who had a history of growth hormone therapy a median of 7 years before the onset of depression. Lastly, second transplant was associated with an increased risk of depression likely reflecting all of the risk factors associated with a first transplant, the increased risk of depression associated with cancer recurrence, and psychological distress from a prolonged period of treatment, greater schooling interruption, and increased stress on family support systems.

Table 2.

Multivariate analysis of risk factors associated with depression

	Depression (n = 194)	Total (n = 1084)	HR (95% CI)	P-value
Age at transplant
0 to < 6 years	33	365	1.0
6 to < 12 years	67	364	2.29 (1.5–3.5)	0.0002
12 to < 18 years	94	355	3.63 (2.3–5.7)	< 0.0001
Donor
Sibling/twin	107	565	1.0
Unrelated	40	269	1.47 (0.9–2.4)	0.13
Other related (parent or related)	29	92	1.97 (1.2–3.2)	0.008
Autologous	18	158	1.53 (0.8–2.8)	0.17
Prednisone for chronic GvHD therapya
No	83	522	1.0
Yes	93	404	1.91 (1.4–2.7)	0.0002
Second transplanta
No	178	940	1.0
Yes	16	144	2.17 (1.2–3.8)	0.006
Obesitya
No	177	957	1.0
Yes	17	127	4.53 (2.7–7.7)	< 0.0001
Growth hormone therapya
No	156	897	1.0
Yes	38	187	1.99 (1.3–3.0)	0.001
Surgery after transplanta
None	124	750	1.0
Orthopedic	31	120	1.76 (1.2–2.7)	0.007
Nonorthopedic	39	214	1.52 (1.1–2.2)	0.03
Gender
Female	93	453	1.0
Male	101	631	0.80 (0.6–1.1)	0.14
Year of transplant
1969–1986	92	348	1.0
1987–1995	61	319	0.97 (0.6–1.5)	0.88
1996–2010	41	417	1.46 (0.8–2.5)	0.18
HLA match
Matched	139	750	1.0
Mismatched	37	176	1.12 (0.7–1.8)	0.61
Acute GvHDa
0–1	87	396	1.0
2–4	89	530	0.80 (0.5–1.2)	0.23
Scleroderma/contracturesa
No	171	1008	1.0
Yes	23	76	1.22 (0.8–2.0)	0.42
Avascular necrosisa
No	170	1007	1.0
Yes	24	77	1.21 (0.7–2.0)	0.47
Pulmonary functiona
Normal/untested	128/42	519/349	1.0
Abnormal	42	216	1.14 (0.8–1.6)	0.49

Abbreviations: CI = confidence interval; HR= hazard ratio.

^aIncluded as a time-dependent variable.

This is the first report of physician-diagnosed depression in a large population of childhood HCT survivors. Depression was found to be a common late effect affecting 12% of current survivors and nearly 40% of patients over 40 years of follow-up. It should be noted that this study comprised a subset of patients diagnosed in clinical practice with depression and due to the limited nature of our follow-up data, we were unable to perform a more comprehensive study of depression and to assess a number of potentially important risk factors and sociodemographic correlates of depression such as employment status, marital status, education and income as this data was not routinely collected, especially decades ago. The most important variables associated with depression were older age at transplant and obesity. Adolescents in particular may need greater psychosocial support. Further research is needed to define at-risk patients, develop assessment tools, and determine beneficial interventions in childhood HCT patients.