The decision-making process in older patients with cancer is challenging. In our study population, the initial cancer treatment plan was deemed inappropriate for 16.7% of patients (n=64). A low MMSE score, malnutrition, and low physical performance were independently associated with change in the initial cancer treatment plan.
In previous studies exploring the impact of geriatric evaluations on treatment decisions in older patients with cancer, the oncologic treatment was modified in 8 to 54% of all patients (with a median of 28%) (14). In our study, the initial cancer treatment plan was changed in only 16.7 % of patients. This difference may be difficult to compare with previous studies because of the populations heterogeneity, the various types of cancer, the various geriatric evaluations and different settings (14). Nevertheless, oncogeriatric evaluation has been implemented in routine clinical practice for a few years in our clinical setting (15), and may have influenced and improved decision-making in this discipline. In addition, geriatric treatment recommendations were closely followed-up by the oncologist when the initial treatment plan was changed (in 91.1% of patients).
In cancer treatment, malnutrition is a substantial parameter to consider, because of its association with treatment toxicity and mortality (28). In our study, malnutrition is significantly associated with changes in planned cancer treatment. Our results are consistent with past studies exploring CGA parameters associated with change in cancer treatment decision. In two studies, a low BMI under 21 kg/m2 was associated with a modification of the cancer treatment plan (29,30), and according to Caillet et al. malnutrition evaluated by MNA, BMI, weight loss or low serum albumin was also associated with changes in cancer treatment (mainly a decrease in treatment intensity) (31). In our analysis, malnutrition was defined only with the MNA score, which may have underestimated the prevalence of malnutrition in this population.
Another CGA parameter significantly linked with change in cancer treatment plan is a MMSE score under 24. Many factors can explain the fact that cognitive impairment may trigger a change in cancer treatment, preferentially from an aggressive treatment to a less aggressive option. First, past studies suggest that older patients with cognitive impairment are less compliant with treatment, which could affect the benefit of chemotherapy for example (32). Secondly, discussion and understanding regarding treatment options may be more complex in this population: this may jeopardize the choice of the treatment. When decision-making capacity is deteriorated, patients tend to choose preferentially the less aggressive option (33). Third there is evidence that chemotherapy can worsen cognitive functions (34). Furthermore cognitive impairment is associated with cancer mortality or the probability of not completing chemotherapy (35–37). But the impact of cognitive impairment or dementia on chemotherapy tolerance, hospitalizations or patient-reported outcome measures remains insufficiently investigated and unclear (38). To our knowledge, only one study concluded that a low MMSE score (<26) was associated with change in cancer treatment plan, specifically in lung cancer (39). The impact of pre-existing cognitive-impairment on cancer-related outcomes needs to be clarified to improve cancer decisions and care in older adults.
In this study, low physical performance defined by a SPPB score less or equal to six, is associated with change in the initial cancer treatment plan. Physical performance tests reflect well frailty in oncogeriatric patients and are easy and rapid to use in clinical settings (11). In previous studies exploring the effect of CGA on treatment decisions, physical performance were not systematically tested or were sometimes limited to the number of falls (29,30,40). According to Farcet et al., the number of Fried’s criteria was significantly associated with a modification of the initial cancer treatment plan (41). In our study, frailty according to the same definition, was not significantly associated with change in treatment plan. As most of the patients were frail or prefail (only 8% were robust), this information is probably not relevant to clinicians. They prefer to base their judgment mainly on the results of the SPPB which seems to offer a better discrimination of subjects with poor physical performance. This is the first time that the SPPB score is identified as a test that could be useful to modify treatment decision in oncogeriatric patients. Its use in routine clinical practice should be considered when evaluating oncogeriatric patients.
This study has some limitations. First, some parameters usually assessed in a geriatric evaluation (such as mood, BMI, or weight loss), were not systematically recorded in our dataset and were not exploited in this analysis. Then, the geriatric assessment was operated in only one hospital, by the same medical team, which may prevent reproducible research in other clinical settings. Moreover, oncologists referred patients to the frailty clinic without using an identified screening tool, but mainly according to their clinical judgment (if the patient seems frail or not). The results of the G-8 assessment show that they referred mainly patients (84%) that needed a CGA. But on the other hand, they may have selected only the frailest patients, and overestimate the robustness of the other patients, who would have been identified as vulnerable with an assessment tool. Indeed, past studies have shown that oncologist’s ability to identify frailty only according to their clinical judgment was low compared to CGA (42,43). This is a potential bias in this study. Finally, a wide spectrum of cancer was included in this work, at different stages, with a wide range of treatments, and with different level of possible complications. It is possible that the decision-making process is different according to the type and stage of cancer, and the type of treatment planned. Specific studies should be planned in the most prevalent cancers and according to the type of treatment to establish specific guidelines in older patient with cancer. This study has also several strengths: this is one of the few studies to evaluate the role of physical performance (SPPB or gait speed) in treatment decision in oncogeriatric patients. We used only international validated tools to assess domains of the CGA, as part of a multidisciplinary evaluation (5), and the median age was relatively high (83 years) (in a previous review evaluating the effect of geriatric evaluation on treatment decisions and outcomes, the median age ranged from 74 to 83 years) (14).