Impact of Comorbidities on Cancer Treatment
Across the selected studies, comorbidities were identified as impacting cancer treatment in a number of ways; however, the causative mechanism of this impact and the degree of impact was neither consistently studied nor reported, making it challenging to draw overall conclusions. The impact of physical comorbidities on cancer treatment and salient findings of each study, as these were statistically analyzed and reported by study authors, is summarized in Tables S3 and S4 of Multimedia Appendix 4 . Major themes included impact of comorbidities on cancer treatment choice, initiation, dose reduction, and other alterations including delay, adverse effects, and discontinuation.
The choice of treatment was noted as affected in some way due to comorbidities in 19 studies [22,24,25,27,28,30-33, 36,38,40,42-44,46-48,50]. Nonstandard treatment and less aggressive treatment were given for older geriatric patients during both primary and secondary treatment regimens, the main factors cited in this being age and physical comorbidities. Hoeben et al [25] reported that the type of chemotherapy had to be modified in 3% out of 57% of patients who received chemotherapy.
Comorbidity affecting treatment initiation was reported in 3 studies [22,25,45]. Hu et al [45] revealed that patients aged 75-79 years were 71% less likely than those aged 65-69 years (OR [odds ratio] 0.29, 95% CI 0.25-0.34) to initiate chemotherapy, and patients with >2 on the comorbidity index were 63% less likely (OR 0.37, 95% CI 0.33-0.42) to initiate chemotherapy after surgery. Age, comorbidity, and marital status were significant predictors for chemotherapy initiation, which showed a model variance of 92.6% in the chi square test. Gross et al [22] studied the presence and absence of individual comorbidities and the initiation of adjuvant therapy. Initiation of therapy for patients with and without coronary heart failure was 36.2% vs 64.9% (OR 0.49, 95% CI 0.40-0.60), with and without chronic obstructive pulmonary disease (COPD) was 55.2% vs 61.5% (OR 0.83, 95% CI 0.70-0.99), and with and without diabetes was 58.3% vs 60.7% (OR 0.81, 95% CI 0.68-0.97). Hoeben et al [25] reported that chemotherapy was not initiated in 43% of patients due to age, comorbidity, or performance status, whereas patient preference accounted for only 17% of noninitiation decisions following surgery [25].
Dose alteration was identified and discussed in 7 studies [23-25,27,29,35,42]. An increase in comorbidities was related to dose modification in patients for ≥2 vs <2 comorbidities and was reported as 40% vs 31% (P<.05) by Goede et al [35] and 59% vs 46% (P=.03) by Klepin et al [23]. Dose reduction was also related to adverse effects from treatment (n=19, 9%) in patients [29]. Hoeben et al [25] reported that 18% and 28% of patients who received chemotherapy underwent alterations in dose and number of sessions, respectively, and in 3% of patients, dose reduction was made before the initiation of treatment. This dose reduction was noted as being not significantly related to age or comorbidity. Jørgensen et al [42] observed that dose reductions in the carboplatin and taxane treatment group in ovarian cancer patients were related to toxicity, but in 17%, it was due to comorbidity or age; however, no significant difference was found based on age for the group receiving only the carboplatin treatment regimen. In rectal cancer, 29.8% of patients had dose reductions (34.3% for 0-1 CCI and 16.7% for >2 CCI; P=.22) [24]. On the contrary, Grønberg et al [27] found no significant differences during therapy and posttherapy in patients (without drug modification) with severe comorbidity.
Treatment delay was examined in 3 studies [25,29,43]. Hoeben et al [25] reported that there was modification in time course between successive chemotherapy sessions in 23% of patients, but this was not related to age or comorbidity. Ferrero et al [43] reported no difference in delay between the age groups 70-75 years and >75 years or based on frailty, but this result was not significant. However, O’Connor et al [29] reported an unplanned delay in treatment for more than a week in about 20% of patients due to toxicity which was significantly related to a history of comorbidities, especially diabetes, hypertension, and low creatinine clearance. An anthracycline-based chemotherapy regimen, CCI ≥1, and hypertension were predictors for treatment delay. A CCI ≥1 was a significant predictor for delay in chemotherapy administration. Age was also a risk factor for delayed treatment.
Treatment discontinuation was reported in 9 studies [22,23,27,29,35,41-43,45]. The most common factors cited in treatment discontinuation were disease progression, toxicity, and patient preference [23,41]. Hu [45] reported that older patients (P<.05) and a <2 comorbidity score (OR 0.63, 95% CI 0.52-0.75) were significant predictors for early discontinuation, and age at diagnosis was the strongest predictor of treatment discontinuation. Similar results were reported by O’Connor et al [29] (OR 4.43, 95% CI 1.55-12.69; P=.045 for >75 years and <75 years). Gross et al [22] found no significant association between individual comorbid conditions and completion of treatment. According to Grønberg et al [27], 69% of patients completed chemotherapy (P=.08); however, the rate was lesser in patients with a severe comorbidity.
The overall response rate for treatment was also found to be lesser in patients with higher comorbidities (75% vs 85% for ≥2 vs <2 comorbidities, respectively; P<.05), but no significant variation was found when the results were adjusted for age and treatment, suggesting that patients with high comorbidity were biased to receiving less vigorous treatment [35]. Ferrero et al [43] reported that complete response to treatment was greater among the 70-75 years age group than among the <75 years age group (60% vs 28.9%, respectively; P=.005). Also, no significant difference between age groups was found for treatment discontinuation due to toxicity in ovarian cancer patients (P=.28) [43]. A similar result was found with respect to the carboplatin-only treatment regimen in a study done by Jørgensen et al [42], whereas in the carboplatin and taxane regimen, performance status and severity of comorbidity were predictors for treatment discontinuation.
Treatment toxicity, adverse effects, or postoperative complications were observed in 14 studies [22-25,27,29,33-35,37,40,41,43]. Goede et al [35] analyzed the individual comorbidities with treatment toxicity and reported no relationship between the variables. However, Grønberg et al [27] observed that the incidence of fever was high in patients with severe comorbidities and also identified that minor comorbidities were not registered in their study, which might have contributed to the result. This suggests the importance of recording the comorbidities, their types, occurrence, and nature in-depth without omitting any details in order to decrease treatment-related adverse effects. In lung cancer, the hematological and nonhematological toxicities were 3% and 24%, respectively [37]. Houterman et al [40] reported no significant difference between treatment complications and comorbidities, irrespective of age. Peters et al [34] reported on recipient site and medical complications out of which the latter was found to be significantly present in head and neck cancer patients with ≥2 comorbidities (OR 2.89, 95% CI 1.71-4.84; P<.001). Phaibulvatanapong et al [41] presented a detailed account of treatment-related complications with adverse effects (grade 3-5) in 83.4% and severe toxicity in 42.4% of patients, both of which were related to performance status in a mixed cancer study population (P<.05). Ferrero et al [43] reported a higher rate of postoperative complications in high-frailty patients compared with low-frailty patients (23.5% vs 4.3%; P=.03). Tan et al [50] similarly reported worse postoperative complications in patients with a CCI >3 or those who had emergency surgery. The study also reported worse perioperative complications and higher death rates among those >85 years old. Hospitalization was not related to congestive heart failure (CHF), COPD, or diabetes, irrespective of whether individuals received treatment [22]. Conversely, Genther and Gourin [33] reported that comorbidities were related to emergency hospital admission (relative risk1.21, 95% CI 1.06-1.38; P=.005) but not to postoperative complications.
Treatment-related toxicity (25.4%, 52%, and 9%) [24] was also another reason cited for treatment discontinuation (1.7%, 15%, and 20%) [23] and dose reduction (29.8%, 51%, and 9%)[29]. Adverse effects varied with the type of treatment (52% vs 41% for those receiving vs not receiving adjuvant chemotherapy, respectively) [25]. O’Connor et al [29] found that history of hypertension is a predictor for poor tolerance of chemotherapy causing treatment delay (OR 2.51, 95% CI 1.02-6.20; P=.046).
Some of the selected studies have noted patients’ personal preference in treatment choice and discontinuation [24,26,32,41,42,50]. For example, Derks et al [32] reported that about 18% of the patients above 80 years of age refused to undergo treatment. Patients diagnosed in more recent years (ie, 2009 or later) were more likely to receive and complete treatment [45,47]. These studies overall show that increased age correlates with an increased likelihood of a patient declining treatment; however, the studies do not identify the specific reasons for this (eg, the impact of comorbidity, impact of function or nonfunction, and so on).
Therefore, several of the selected studies show a strong association between comorbidities and treatment dose alteration, noninitiation of treatment, treatment choice, and early discontinuation of treatment. Due to significant variation in cancer types or sites, patient cohorts, recording of comorbidities, and several other variables, it is, however, difficult to draw clear conclusions regarding the influence of comorbidities in particular on the treatment decisions and the effects among the broader patient population and that of older cancer patients in particular.