An assessment of the use of patient reported outcome measurements (PROMs) in cancers of the pelvic abdominal cavity: identifying oncologic benefit and an evidence-practice gap in routine clinical practice

Of the seven RCTs included [14‐20], two were interventions implemented in pancreatic cancer populations [15, 16], two were undertaken in a colorectal cancer setting [14, 17], one in bladder cancer [19], one in gastric cancer [18] and one in prostate cancer [20]. Two studies were based on the same RCT (Clinical Outcomes Surgical Therapy trial NCCTG 93-46-53), but investigated the prognostic significance of measuring baseline patient reported outcomes (PROs) in slightly different capacities [14, 17]. These two studies utilised results from a surgical RCT where the intervention compared open versus laparoscopic techniques in terms of post-operative complications and patient outcomes. Both studies used the Symptom Distress Scale and QOL Index recorded preoperatively to demonstrate the significant prognostic impact of such quality of life (QoL) measurements on the overall survival (OS) of colorectal cancer patients, and to demonstrate that these measures are more sensitive than clinician reported outcomes (CROs) in predicting mortality. Of note, longer term combined analysis of patients in both arms of the intervention identified lower patient baseline outlook as associated with decreased overall survival.

Four RCTs were secondary analyses of chemotherapy interventions whereby pancreatic, bladder and gastric cancer patients were randomised to receive varying regimens of cytotoxic therapy [15, 16, 18, 19]. Noteworthy observations from these studies included the prognostic significance of pain and fatigue as independent indicators for survival in pancreatic cancer, although these measures were found to be less prognostic than the carbohydrate antigen 19-9 (CA-19) [15]. In addition, this study surmised that QoL did not predict tumour response to chemotherapy.

A second RCT for pancreatic cancer patients utilised the EORTC QLQ-C30 questionnaire to identify no significant difference in QoL in treatment arms over time [16]. Subsequent multivariate analysis identified physical functioning, constipation and dyspnoea as significant prognostic factors in this patient cohort, with severely impaired physical functioning imparting the strongest negative effect on overall survival [16]. In the context of the bladder cancer RCT, Roychowdhury et al. also utilised the EORTC QLQ-C30 questionnaire to demonstrate physical functioning as a significant and independent prognosticator for time to event endpoints [19]. This RCT also identified a potential prognostic role for fatigue and anorexia in both treatment arms. Interestingly, in univariate analysis, higher role functioning was identified as a positive prognostic factor but, paradoxically, in the multivariate model longer overall survival was associated with lower role functioning.

In the setting of gastric cancer, three RCTs assessing fluorouracil-based combination chemotherapy were combined to investigate whether pretreatment QoL predicts survival in patients with locally advanced or metastatic disease [18]. In this study, and similarly to the other RCTs, better physical and role functioning predicted increased survival. Data was collated using the EORTC QLQ-C30 instrument and suggested the role of global QoL score as a strong prognostic factor.

In contrast to the other RCTs, a European hormonal therapy/radiotherapy intervention in prostate cancer identified that HRQoL factors, as measured by the EORTC QLQ-C30 instrument, did not accurately predict overall survival once clinical and biochemical factors were accounted for [20]. Despite baseline global health status being associated with overall survival in other tumour types, this relationship was not demonstrated in this study.

Full details of the 11 included cohort studies are contained in Table 2. One was in the context of aggregated RCT data [21] and 10 were original cohorts [22‐31]. The tumour breakdown of these patient cohorts were as follows: five analyses included urological cancer patients [21‐23, 27, 28], four colorectal [21, 22, 30, 31], three gynaecological [21, 22, 26], three pancreatic [24, 25, 29] and one gastric cancer cohort [31].

An analysis by Jayadevappa et al. of 318 younger prostate cancer patients identified that low risk biochemical recurrence is mostly indicative of better generic and prostate specific HRQoL [28]. Similarly, an American analysis of 917 prostate cancer patients demonstrated that patient self-rated health is a potential confounder in the relationship between patient satisfaction and survival; thus suggesting that future studies investigating patient satisfaction should include collection of self-rated health modules [27]. In renal cancer, a study by Graham et al. observed that baseline use of the Edmonton Symptom Assessment System provides a modest degree of prognostic information about survival, independently of other widely used prognostic models [23]. These results were consistent with previously reported data examining the prognostic function of the FACT-KSI instrument.

In pancreatic cancer patient cohorts, analyses further suggested a role for the EORTC QLQ-C30 instrument to provide prognostic information for survival [29]. As reported above, baseline global health was an independent prognosticator and, interestingly, the study found that the probability of survival increased significantly if cognitive function improved within three months of treatment [29]. Of note, an analysis of 66 Norwegian pancreatic cancer patients also demonstrated that cognitive function, as measured by the Edmonton Symptom Scale, was an independent prognostic factor [24].

In the context of colorectal cancer, a large UK study of 501 patients identified that patients with high baseline global QoL have a 1-year survival that is almost double that of patients with a score below median value [30]. Multivariate analysis demonstrated that symptom and functioning measures, as recorded using the EORTC QLQ-C30, appeared to be a stronger predictor of overall survival compared to clinician measured performance status. A smaller analysis of 47 patients with advanced gastric or colorectal cancers identified that overall physical condition and global QoL was an independent prognosticator of overall survival [31]. Indeed, further preliminary analyses demonstrated an association between psychological response to cancer and survival.

A Danish analysis of a gynaecological patient cohort investigated patient quality of life and satisfaction as an alternate avenue for exploring the consequences of diagnostic delay [26]. This study identified that, in ovarian cancer patients specifically, pain was associated with reduced overall survival. In a subset of the cohort with endometrial cancer, a number of QoL domains including overall QoL, physical, emotional and role functioning, nausea and vomiting, pain, dyspnoea, and appetite loss were independently associated with survival [26].

Three studies recruited prostate and pancreatic cancer patients consecutively [32‐34]. In the context of radiotherapy, overall survival and disease-free survival in prostate cancer patients with localised disease were predicted by socioeconomic status, psychological factors and patient self-reported QoL [32]. In this study, different QoL domains demonstrated favourable or unfavourable impact; patients with reports of few or no physical complaints predicted shorter survival whereas reported pain was prognostic for longer overall survival.

A case series of 55 pancreatic cancer patients identified a borderline significant association between baseline health and physical measures and survival after adjustment for disease stage at diagnosis [33]. Indeed, this study also suggested that patient satisfaction with QoL provides useful prognostic information.