Syrjala KL, Langer SL, Abrams JR, Storer B, Sanders JE, Flowers MED, Martin PJ. Recovery and Long-term Function After Hematopoietic Cell Transplantation for Leukemia or Lymphoma. JAMA. 2004;291(19):2335-2343. doi:10.1001/jama.291.19.2335
Author Affiliations: Clinical Research Division, Fred Hutchinson Cancer Research Center (Drs Syrjala, Langer, Abrams, Storer, Sanders, Flowers, and Martin) and the Departments of Psychiatry and Behavioral Sciences (Drs Syrjala, Langer, and Abrams), Pediatrics (Dr Sanders), and Medicine (Drs Flowers and Martin), University of Washington School of Medicine, Seattle..
Context Hematopoietic cell transplantation (HCT) is an effective and widely
used treatment for hematologic malignancies. The rate and predictors of physical
and emotional recovery after HCT have not been adequately defined in prospective
Objective To examine the course of recovery and return to work after HCT.
Design, Setting, and Patients Prospective, longitudinal cohort study at a US academic center specializing
in HCT. Function was assessed from pretransplantation to 5-year follow-up
for 319 adults who had myeloablative HCT for treatment of leukemia or lymphoma
and spoke English. Of the 99 long-term survivors who had no recurrent malignancy,
94 completed 5-year follow-up.
Main Outcome Measures Physical limitations, return to work, depression, and distress related
to treatment or disease were evaluated before transplantation, at 90 days,
and at 1, 3, and 5 years after HCT.
Results Physical recovery occurred earlier than psychological or work recovery.
Only 21 patients (19%) recovered on all outcomes at 1 year. The proportion
without major limitations increased to 63% (n = 57) by 5 years. Among survivors
without recurrent malignancy, 84% (n = 74) returned to full-time work by 5
years. Patients with slower physical recovery had higher medical risk and
were more depressed before HCT (P≤.001). Patients
with chronic graft-vs-host disease (P = .01), with
less social support before HCT (P = .001), and women
(P<.001) were more depressed after transplantation.
Transplant-related distress was slower to recover for allogeneic transplant
recipients and those with less social support before HCT (P≤.01). Patients who had more experience with cancer treatment before
beginning HCT had more rapid recovery from depression (P = .04) and treatment-related distress (P =
Conclusions Full recovery after HCT is a 3- to 5-year process. Recovery might be
accelerated by more effective interventions to increase work-related capabilities,
improve social support, and manage depression.
Survival rates after myeloablative hematopoietic cell transplantation
(HCT) have improved considerably during the past 30 years.1 The
growing population of patients who have survived this treatment has intensified
the need to determine the course of recovery and to identify risk factors
for less favorable functional outcomes. Improved understanding of recovery
will facilitate more accurate informed consent, permit better planning by
patients, families, and medical teams, and enable the design of interventions
to improve functional recovery.
Medical complications of HCT have been well documented and include chronic
graft-vs-host disease (GVHD), recurrent infection, pulmonary complications,
premature menopause, cataracts, osteoporosis, avascular necrosis, infertility,
recurrent malignancy, and secondary malignancy.2- 10 Cross-sectional
studies have documented functional and mental health from 1 to 10 years after
transplantation but have not described the course of recovery or identified
pretransplantation factors that predict recovery.11- 19 Other
studies have been limited by lack of pretransplantation baseline assessment,20 shorter follow-up,21- 25 or
small sample size when follow-up extends to 3 years.26,27
We conducted a prospective, longitudinal study to examine recovery of
physical and mental health and return to work after HCT for treatment of leukemia
or lymphoma. Assessments began before the transplantation and continued for
5 years to identify pretransplantation characteristics associated with outcomes
after HCT. Based on previous research, we hypothesized that physical and emotional
recovery would occur by 1 year, autologous transplantation recipients would
recover more rapidly than allogeneic transplant recipients, and risk factors
for slower or poorer recovery would include higher medical risk as indicated
by relapsed disease at the time of transplantation, history of extensive chronic
GVHD, and pretransplantation depression.
A total of 335 adults who were preparing for a first myeloablative HCT
for leukemia or lymphoma were invited to participate in this study. Eligible
participants were at least 18 years old, had sufficient English language proficiency
to complete the assessments, and were available to complete the baseline assessment
in the ambulatory clinic before the transplantation. Each patient provided
written informed consent. Sixteen patients declined to participate. The transplant
center institutional review board and scientific review committee approved
the study procedures, the consent form, and assessments. Pretransplantation,
90-day, and 1-year self-administered assessments were conducted in the ambulatory
clinic with a researcher available to answer questions. The 3- and 5-year
repeated assessments were conducted by mail with a researcher available by
toll-free telephone line for questions. The researcher called to follow up
with participants who did not return forms within 3 weeks. All participants
were contacted for follow-up until 5 years, death, or the development of recurrent
malignancy, whichever occurred first.
Primary Outcomes. Three self-report instruments
were used to evaluate outcomes: the ambulation subscale of the Sickness Impact
Profile (SIP)28 as a measure of physical limitations,
the Beck Depression Inventory (BDI)29 as a
measure of clinical depression, and Cancer Treatment Distress scale30 as a measure of distress or worry specific to the
transplantation and associated complications. At each time point, participants
defined their work situation. At 3 and 5 years, participants were asked to
provide a date of return to work, which was operationally defined as either
full-time work outside the home, full-time school, or part-time school coupled
with part-time work.
In this sample, the SIP ambulation subscale correlated highly with the
entire SIP physical function subscale across time (r =
0.74 to 0.91) and encompassed the major areas of long-term difficulty for
transplant recipients over time. As a result, the shorter ambulation subscale
was used to assess physical limitations. Continuous SIP scores were used for
descriptive analyses, but because the distribution showed strong positive
skewing after 90 days (1-year mean = 2.53; range, 0-37.41), we used a dichotomous
score for statistical models. Patients with scores more than 1 SD above age-
and sex-adjusted population norms were considered to have clinically meaningful
Higher scores on the BDI indicated greater severity of depression. General
population norms29 were used to define categories
of no depression (<10), mild depression (10-15), moderate depression (16-21),
and severe depression (>21).
For the Cancer Treatment Distress measure, 27 items such as "nausea
or vomiting," "possibility of relapse," and "being a burden to other people"
were rated for the extent to which they caused distress or worry (0 = no distress,
3 = severe distress). A mean score was calculated with a possible range of
0 to 3. This distress measure has predicted pain, nausea, and distress among
transplant recipients during treatment better than generalized measures of
anxiety or depression.30,31
Medical Records Extraction of Risk Factors. Medical
data were analyzed as risk factors for outcomes and were collected from the
clinical research database, including diagnosis, conditioning regimen, history
of treatment before HCT, stem cell donor (autologous vs allogeneic related
or unrelated), date of onset of clinical extensive chronic GVHD, and dates
of recurrent malignancy and death. Pretransplantation risk categories were
assigned with 3 levels depending on diagnosis and disease status.32 Chronic myeloid leukemia in chronic phase was classified
as low risk, acute leukemia or lymphoma in remission and chronic myeloid leukemia
in accelerated phase were classified as moderate risk, and acute leukemia
or lymphoma in relapse and chronic myeloid leukemia in blast crisis were classified
as high risk.
Self-reported Risk Factors. Income (≤$49 999
vs ≥$50 000), marital status, and education before HCT (<4-year
college degree vs ≥4-year college degree) were treated as dichotomous variables.
The Social Support Inventory33 was administered
at baseline as a possible predictor of outcomes. This instrument was not administered
at the posttransplantation time points simply to minimize patient burden.
Scores reflected overall satisfaction with multiple kinds of support (advice,
major and minor assistance, love, encouragement, and understanding). Ratings
were made on 7-point scales for each type of support, with higher numbers
indicating greater satisfaction.
Descriptive and inferential analyses were performed using SPSS version
10.0 (SPSS Inc, Chicago, Ill) and Stata version 6.0 (Stata Corp, College Station,
Tex). To compare recovery rates across the 3 primary outcomes, scores for
physical function, depression, and treatment distress were standardized to
the same scale using z scores derived from the entire
data set. Generalized estimating equations were used to test for differences
between trajectories of these outcomes. This analysis method accounted for
the statistical dependence of repeated observations over time and permitted
inclusion of results for participants who did not complete all assessments.34 Generalized estimating equations were also used to
develop statistical models of association between baseline characteristics
and outcomes, using models that included time, risk factor, and time ×
risk factor interaction terms. The model for physical function used the dichotomous
version of that variable and assumed correlated binomial errors; models for
depression and distress assumed correlated normal errors.
All models tested the same 15 risk factors: stem cell donor (autologous
vs allogeneic), medical risk category, clinical extensive chronic GVHD, diagnosis,
total body irradiation in the conditioning regimen, history of radiation therapy
before transplantation, history of chemotherapy before transplantation, age,
sex, education, income, marital status, and pretransplantation levels of physical
limits, depression, and treatment distress. A parsimonious model for each
outcome was calculated via forward stepwise entry to the model for risk factors
with either a main effect or interaction with time significant at the P<.05 level. Cumulative incidence and Cox proportional
hazards estimates were used to examine return to full-time work using the
same risk factors. These analyses were conducted with the subsample of patients
who had a history of work or school outside the home prior to transplantation.
Figure 1 displays the consenting
and follow-up of eligible patients. The lower rates of response at 90 days
and 1 year resulted from the fact that assessments were conducted at the transplant
center with those on site for follow-up, whereas the 3- and 5-year assessments
were administered by mail to all survivors without recurrent malignancy. Among
patients who completed the 1-year assessment, residual treatment distress
was lower for those who completed the 90-day assessment than for those who
did not (P = .001), but the 2 groups had no significant
differences in physical limitations or depression. Among survivors who completed
the 3-year assessment, no significant differences in any of the outcome measures
were detected between those who completed the 90-day assessment and those
who did not.
Table 1 summarizes patient
demographic and clinical characteristics as a function of cohort: those in
the cohort consenting to participate (n = 319), those not consenting to participate
(n = 317), and those consenting and in the subgroup surviving to 5 years without
recurrent malignancy (n = 99 of the 319 patients). Based on comparisons between
those consenting vs not consenting to participate and between those consenting
and the subgroup who survived to 5 years, these groups did not differ in most
characteristics. Five-year survival was 33% in the consenting sample and 21%
in the not consenting sample (log-rank P<.001).
Patients not consenting vs consenting were more likely to have diagnoses of
acute leukemia in relapse (P<.001) and were more
likely to be high risk (P<.001). The cohort not
consenting had a higher number of nonwhite or Hispanic patients (P = .004). Among patients not consenting, 12 nonwhite or Hispanic patients
were not eligible because they did not speak English; the cohort difference
was not significant when these patients were excluded from the analysis.
As expected, the 5-year survivor cohort was generally medically healthier
than the cohort of all consenting patients. The cohort of all consenting patients
more often had previous radiation therapy (P = .03),
were more likely to be in relapse before transplantation (P = .001), and were more frequently at high medical risk (P<.001) than the subgroup of 5-year survivors. In comparing outcomes
between all consenting patients and those who were 5-year recurrence-free
survivors, the 2 groups had few differences (Table 1).
The trajectories of physical limitations, depression, and treatment
distress varied over time and differed from each other (P<.001) in generalized estimating equation analysis (Figure 2). Physical limitations reached a peak at 90 days after
HCT, followed by improvement at 1 year and no further significant change at
3 and 5 years. Before the transplantation, 25% of patients (n = 81) reported
major physical limitations. The proportion increased to 44% (n = 76) of survivors
at 90 days and then decreased to 12% (n = 14) at 1 year. At 3 and 5 years,
respectively, 22% and 18% of survivors (n = 23 and 16) reported major physical
limitations. Treatment distress was high before HCT and at 90 days after the
transplantation and then decreased steadily at each subsequent assessment.
Depression changed less dramatically over time. Mean scores at 90 days were
unchanged from baseline, while scores at 1, 3, and 5 years were lower, with
no change after the first year. Thus, both physical limits and depression
analyses confirmed our first hypothesis that most recovery would occur by
1 year, but treatment distress did not recover in this time frame and continued
to resolve over 3 to 5 years.
Because averages can be deceiving, we defined clinically meaningful
rates of distress and depression for further analysis. We used the 5-year
mean score plus 1 SD as the threshold for clinically meaningful distress.
According to this standard, 79% of survivors (n = 84) remained distressed
at 1 year, 42% (n = 35) at 3 years, and 13% (n = 11) at 5 years. We used general
population norms to define the threshold for depression in this sample. At
90 days, 33% (n = 102) reported depressive symptoms, whereas at 5 years 19%
(n = 18) had depressive symptoms. Most of those with symptoms had mild depression.
At each assessment, 7% to 11% of patients had moderate to severe depression.
Among the 188 patients who had at least 2 assessments, 78% (n = 146) never
had moderate to severe depression, 21% (n = 39) had moderate to severe depression
at some (but not all) assessments, and 1% (n = 2) had moderate to severe depression
at all assessments. An additional 31% (n = 58) were never moderately or severely
depressed, but had mild depressive symptoms at 1 or more assessments.
Analyses only partially confirmed that risk factors for poorer or slower
recovery would include allogeneic transplant, higher medical risk, clinical
extensive chronic GVHD, and depression. Risk factors for physical limits included
depression, higher medical risk, and a history of radiation therapy before
transplantation (Table 2). Before
HCT, physical limitations were reported more frequently among patients in
the moderate or high medical risk categories than among those in the low-risk
category. After transplantation and long-term, the groups were similar in
rates of patients reporting physical limits and different from their relative
rates of impairment before transplantation (Figure 3). Throughout follow-up, physical limitations occurred more
often among patients who were depressed before HCT than among those who were
not depressed (Figure 4). A history
of radiation therapy before transplantation also increased the risk for long-term
physical limitations. Contrary to our hypotheses, neither allogeneic transplant
nor chronic GVHD predicted physical limits.
Patients with chronic GVHD and women reported more depression. Similarly,
those with no history of chemotherapy before their transplantation and with
lower satisfaction with support before HCT reported more depression across
time (Table 3). If patients had
physical limits before HCT, they were more depressed before transplantation
but not afterward. At 90 days, depression had declined more in patients with
a history of radiation therapy before starting HCT treatment than in patients
without previous radiation therapy. In relation to our hypotheses, analyses
confirmed that patients with chronic GVHD reported more depression. However,
contrary to our hypotheses, there was no difference in depression between
allogeneic and autologous transplant recipients or between those with higher
or lower medical risk.
Risk for treatment-related distress was greater over time for patients
who had lower satisfaction with support before the HCT, those who had no history
of chemotherapy before they received transplant chemotherapy, and in acute
leukemia patients vs those with chronic leukemia or lymphoma (Table 4). Confirming our second hypothesis, distress declined more
rapidly between pretransplantation and 3 years in patients who received autologous
vs allogeneic stem cells, although the pattern of recovery from distress was
otherwise similar (Figure 5. Women
reported more distress before HCT and remained at a higher rate than men until
they declined more rapidly in distress between 3 and 5 years. Patients with
a college or postgraduate education were slower to decline in distress. Contrary
to our third hypothesis, medical risk and extensive chronic GVHD were unrelated
to treatment distress within or across time.
Other risk factors tested did not significantly predict the primary
outcomes. Factors unrelated to any of the outcomes tested included age, income,
marital status, and whether a patient received total body irradiation as part
of the treatment regimen. Cohort diversity was not adequate to consider race
or ethnicity as a predictor of outcomes.
Eighteen of the 281 patients who had a history of school or work outside
the home and who survived past 1 year without recurrent malignancy were excluded
from analyses of return to work because they did not report return to work
information. The 146 patients who died or had recurrent malignancy before
1 year were scored as not returning to work. Figure 6 displays the cumulative incidence of return to full-time
work, with competing risk of death or recurrent malignancy also indicated.
Among patients with a history of work outside the home, 20% returned to full-time
work by 1 year, 31% by 2 years, 33% by 3 years, and 34% by 5 years. Among
5-year survivors without recurrent malignancy and with a history of work or
school outside the home before transplantation, 84% returned to full-time
work or school. Return to work was significantly delayed for women (hazard
ratio, 0.52; 95% confidence interval [CI], 0.33-0.82; P = .005). Patients with extensive chronic GVHD also had a nonsignificant
trend toward delayed return to work (hazard ratio, 0.70; 95% CI, 0.4-1.1; P = .10). Cumulative incidence of return to work at 1 year
for patients receiving autologous transplants was 23% and for allogeneic recipients
it was 19%, a nonsignificant difference (hazard ratio over the first year,
0.57; 95% CI, 0.3-1.1; P = .13). The other medical,
demographic, psychological, and physical risk factors tested with all outcomes
did not predict return to work.
To determine rates of full recovery, we evaluated the number of impairments
reported at 1, 3, and 5 years. This analysis included physical and work limitations,
depression, and distress. Among respondents at each time point, 19% (n = 21)
reported no impairments at 1 year, 49% (n = 51) reported none at 3 years,
and 63% (n = 57) reported none at 5 years. At 5 years another 25% (n = 23)
had a single area of impairment, while 7% (n = 6) had 2 areas of impairment.
Results of this prospective longitudinal study show that recovery after
HCT occurs gradually over 1 to 5 years as measured by improvement in physical
function, return to work, depression, and treatment-related distress. Given
adequate time, 84% of survivors returned to full-time work. At some point
during treatment or recovery, 22% of the patients had symptoms consistent
with clinical depression while an additional 31% had mild depressive symptoms.
Higher levels of depression, lower levels of physical function, and less satisfaction
with social support before HCT increased the risk of impaired physical and
emotional recovery after the transplantation. Women had increased risk for
depression, treatment-related distress, and delayed return to full-time work.
Conversely, previous experience with chemotherapy or radiation therapy before
beginning HCT seemed to facilitate recovery from the psychological aspects
of this intensive treatment. Extensive chronic GVHD delayed recovery from
depressive symptoms and marginally delayed return to work but did not significantly
influence other outcomes. Although clinical impression suggests that autologous
transplant patients recover more rapidly than allogeneic recipients, these
groups differed only in more rapid recovery from treatment-related distress
for autologous stem cell recipients.
Our results emphasize that recovery is not a unidirectional process
of improvement. In general, physical recovery occurred earlier than emotional
recovery. However, physical function remained susceptible to disruption by
a variety of medical complications that can occur after the transplantation.14,15,35 The proportion of
participants with physical limitations increased from 12% at 1 year to 18%
to 22% at 3 and 5 years, respectively. Similarly, emotional recovery fluctuated.
Eight percent of patients who were not clinically depressed at 1 year were
clinically depressed at 3 years. Although depression was not more prevalent
in patients with physical problems at 1 year, at 3 and 5 years survivors with
residual physical limitations were more likely to be depressed. Thus it seems
that continuing physical complications began to take a toll on long-term survivors.
Both physical and psychological recovery progressed more slowly among
participants with depression before HCT. In other studies, depression has
been associated with increased mortality after the transplantation.36 While our analyses did not examine the relationship
between depression and mortality, we found that moderate or severe depression
and treatment-related distress adversely affected quality-of-life outcomes.
Of note, depression and distress differed in both their trajectory of recovery
and in risk factors, which emphasizes the value of separately evaluating these
syndromes. Also important, the seeming advantages of lack of previous treatment
and more education delayed recovery from treatment-related distress rather
than being protective. Adequate management after a complex medical intervention
such as HCT requires routine screening to detect depression or distress and
allocation of resources for treatment. Cost-effective methods to achieve these
aims are readily available.37,38
In a positive finding, medical risk factors that predict survival and
short-term quality of life19,20,24 did
not have detectable influence on long-term outcomes among patients who survived.
Among patients alive at 3 and 5 years, those who were older, those receiving
allogeneic transplants or total body irradiation, and those with higher-risk
diagnoses and disease stages had the same long-term quality of life as did
patients with better medical prognoses. The exception, as hypothesized, was
that the continuing presence of chronic GVHD increased the risk of long-term
depression and was marginally associated with later return to work. Anecdotal
reports from patients suggested that return to work was driven by financial
needs or facilitated by flexibility in work responsibilities and did not necessarily
indicate complete medical recovery.
Five-year survivors without recurrent malignancy did not differ from
the entire cohort assessed in regard to physical limits or treatment-related
distress after transplantation. Rates of clinical depression differed only
at 1 year after HCT. These findings indicate that the changes described are
not a result of more impaired patients dropping out of the longitudinal assessment
as they died or their disease recurred. Rather, individual patients, whether
long-term survivors or not, had similar patterns of change in physical limits,
depression. and treatment-related distress.
As in all long-term follow-up studies, biases might have been introduced
by missing data or unbalanced representation of the population. Groups that
are underrepresented in our sample include autologous transplant recipients,
patients who did not receive total body irradiation, and nonwhite or Hispanic
transplant recipients. Our study did not include patients who were hospitalized
soon after they arrived at the center. These patients were more likely to
have high-risk characteristics and higher rates of mortality. However, the
study did include an adequate representation of high-risk patients to permit
statistical control for this factor. Our study also did not provide follow-up
for individuals who had survived recurrent malignancy after the transplantation.
These results are both encouraging and cautionary. Patients, families,
and medical teams depend on accurate recovery data when planning for posttransplant
needs. Expectations that contradict actual experience cause stress for survivors
and potential conflicts with family, work, and the medical team. To facilitate
realistic planning, clinicians and patients should understand that full recovery
requires more than a year for most survivors. Patients at risk for delayed
recovery can be identified before transplantation. Rehabilitation programs,
similar to those that have accelerated recovery for cardiac patients,39 might improve the physical and psychological health
of HCT recipients and other patients who have survived after curative treatment