The impact of comorbid disease history on all-cause and cancer-specific mortality in myeloid leukemia and myeloma – a Swedish population-based study

In this large population-based study, using prospectively recorded information on
comorbid diseases, we showed that patients with a history of comorbidity at diagnosis
of AML, CML or myeloma had a higher all-cause but also cancer-specific mortality compared
with patients without such history, reflecting an impact on disease-specific outcome
in these malignancies. Renal disorders were associated with a markedly higher cancer-specific
mortality among all three patient groups (but were uncommon, prevalence 0.5–1 %).
Cerebrovascular disease, dementia and psychiatric disease were associated with an
increased risk of cancer-specific death in AML and myeloma patients, liver and rheumatologic
disease increased risk in AML only, and cardiovascular and chronic pulmonary disease
in myeloma only. In absolute terms, the 5-year probability of cancer-specific death
was greater than that of other-cause death among all patients aged 60–89 years except
among male patients ?70 years with CML. Comorbidity contributed most to cancer-specific
death among younger patients (?70 years) in AML, whereas the impact was constant
by age in myeloma.

AML

The achievement of complete remission and long-term survival in AML mostly requires
intensive combination chemotherapy, and outcomes are strongly dependent upon age and
performance status 34], 35]. During the study period in Sweden, the majority of the patients diagnosed up to
the age of 80 years received intense treatment 34]. Whether further prognostic stratification and personalized therapy can be achieved
by adding a more systematic evaluation of comorbidity has been investigated in a few
previous studies. In most 12], 13], 36]–38], but not all 14] of these, comorbidity assessed using the Charlson index was independently associated
with a worse overall survival. Etienne et al. (N?=?133) showed that comorbid diseases (with an index score ?1) negatively predicted
complete remission rate 13]. In two recent large studies, a lower likelihood of treatment with intense chemotherapy
was noted in the presence of comorbid disease 14], 37]. In Ostgard et al. 14], comorbidity was not associated with survival when adjusted for performance status
and other factors. Performance status could however be considered an intermediate
explanatory factor rather than a true confounder and therefore an association between
comorbidity and survival through lower performance status is still plausible. In that
study, outcome was also investigated in relation to specific comorbid disease, and
dementia, heart failure and renal failure were associated with opting-out of intensive
therapy 14]. This is in line with our findings of a decreased cancer-specific survival in patients
with renal disease and dementia. A similar explanation is plausible among patients
with cerebrovascular and psychiatric disease, also noted to have a worse cancer-specific
survival in our study.

Ostgard et al. also observed an indication of a stronger association between comorbidity
and outcome among patients ?60 versus ?60 years of age. Extending these previous
results, we show a clear differential effect of comorbidity by age with a larger prognostic
importance of comorbidity among patients 60–69 years versus older patients. Hence,
our results provide additional support for the notion that poor outcomes among AML
patients ?70 years cannot be explained solely on the basis of increased prevalence
of comorbidity by age but 14], 34], rather through a more general low treatment tolerance at older ages.

CML

CML survival has improved greatly with the introduction of tyrosine kinase inhibitors
such as imatinib (introduced in 2001 in Sweden) 39] although elderly Swedish CML patients (?79 years) still have a 5-year relative survival
of only 60 % 40], that may reflect under-implementation of tyrosine kinase inhibitor use 39], 40]. A few previous studies have shown a negative impact of comorbidity on CML survival
in line with our results, mainly reflected in a poorer event-free survival 36] or lower degree of complete cytogenic response 41], 42]. Previous studies have assessed comorbidity through pooled indices including the
Charlson comorbidity index 36], 41], or the adult comorbidity evaluation-27 score and cumulative illness rating scale
36], 41], but have not investigated survival by specific comorbidities, perhaps due to low
numbers. We show for the first time that prior renal disease is associated with a
poorer cancer-specific survival in CML. Renal disease is not an absolute contraindication
for use of tyrosine kinase inhibitors, but glomerular filtration rate may decrease
further during tyrosine kinase inhibitors treatment 43]. Thus, dose reductions 44] or caution to prescribe tyrosine kinase inhibitors could potentially explain this
finding. Also, high comorbidity index has been associated with an increased risk of
toxicity to tyrosine kinase inhibitors in two previous studies 41], 45].

A previous cohort study indicated that treatment of elderly CML patients (n?=?181, median age 79 years) might be influenced by the individual physician’s perception
and could be improved by utilizing comorbidity indices 36]. In our study, comorbidities were only associated with a higher probability of CML-specific
death among men 60–69 years of age but not among older patients. Among the elderly
males (?70 years of age), other-cause deaths outweighed CML-specific deaths regardless
of comorbidity. In contrast, among women with CML, comorbidity was only associated
with a higher probability of CML-specific death in the oldest group (80+). Women with
CML were more likely to die of CML-specific rather than other-cause death up to 89 years.
Previous Swedish studies have noted a possible reluctance to treat elderly patients
with tyrosine kinase inhibitors during the investigated time-period 39], 46]. The present findings indicate that CML outcome could potentially be further improved
among elderly, especially female patients.

Myeloma

Survival in multiple myeloma has increased during recent decades especially among
younger patients (?60-70 years), likely due to a combination of factors including
increasing use of high-dose Melphalan with stem cell support and thalidomide as well
as improvements in supportive care 47], 48]. Previous studies have reported comorbidities to be of critical prognostic importance
at myeloma diagnosis using different comorbidity indices 49], 50]. In particular, renal impairment (pre-existing or disease-related) has been identified
as an important determinant for myeloma outcome 50], 51]. Kleber et al. have developed the Freiburg comorbidity index including performance
status, renal impairment and lung disease, and have reported large differences in
overall survival among 466 myeloma patients (median age 62 years) by the presence
or absence of a combination of these factors 49]. In our study including ~?4,500 myeloma patients with a median age of 72 years, we
confirm the adverse prognostic impact of pre-existing renal and pulmonary disease,
and extend the list of disorders associated with a higher risk of cancer-specific
death to also include cardiovascular and cerebrovascular disease, dementia and psychiatric
disorders. Hence, our study suggests that future evaluations of comorbidity and myeloma
outcome in larger studies may benefit from including a broader list of disorders 52]. Interestingly, and in contrast with AML and CML, the prognostic impact of comorbid
disease seemed relatively constant by age (among patients aged 60–89 years) 53].

The strengths of our study include the large size of the cohort, the high quality
and coverage of the registers used as well as the population-based unselected setting,
evaluating the effect of 12 severe comorbid diseases on outcome of hematological malignancies
in the most recent decade. We also, for the first time in this setting, used a novel
methodology to estimate probabilities of death associated with comorbidities in patients
with hematological malignancies, in the presence of competing risks. While traditional
ratio estimates of net survival (such as those presented in Table 2) are important to identify and evaluate the impact of prognostic factors associated
with the disease under study, competing risks analyses may provide additional insights
to understanding the real-world prognosis of the patients. This is because, in contrast
to estimates of net survival, a competing risks analysis takes into account that causes
other than the malignancy may kill the patient first and thereby preclude death from
the malignancy. Thus, competing risks analyses more appropriately reflect the absolute
impact of a prognostic factor on prognosis 54]. The advantage of studying the three different hematological malignancies together
was to contrast between malignancy types needing intensive treatment upfront (AML)
and those with slower tumor progression in need of more intermediate-intensity treatment
(myeloma, CML). An important limitation of our study was the lack of clinical data
such as performance status, disease-specific prognostic determinants including genetic
abnormalities, and treatment. Since prior comorbidity may lead to lower performance
status and may affect choices of treatment, rather than the other way around, performance
status and treatment intensity may be considered explanatory factors rather than true
confounders when estimating the impact of comorbidity on survival. Another limitation
to consider is the definition of deaths as cancer-specific. Although the accuracy
of the classification of main underlying cause of death has been found to be high
for malignant diseases in the Swedish Cause-of-death registers 30], 32], some leukemia/myeloma deaths may have been erroneously classified as non-cancer-related
or vice versa. However, given that the majority of the deaths were cancer-specific
and that we also present patterns of all-cause and other-cause deaths, a minor degree
of such misclassification does not threaten our main conclusions.