As we have established the general patterns of QOL after the three major treatments for localized disease, the question remains: can we properly compare the modalities to each other? As stated, no randomized trials are available and, therefore, retrospective and observational cohort studies must be interpreted, understanding the inherent weaknesses in these study designs. In addition, there are subtle points of technique that one can look for in the methods section that may shed light on the validity of the conclusions.
1. Pretreatment data. In many cases, the decision to analyze quality of life came after treatment or after suitable QOL scales were available. Therefore, authors must either omit pretreatment data, compare to published norms or ask patients to recall their pretreatment function. As mentioned above, this issue is controversial and it is not established which method is preferred.
2. Selection bias. As all QOL comparisons are non-controlled, there will be selection biases that affect age, comorbidity and probably pretreatment erectile/urinary function. Thus, these QOL scores must be statistically adjusted, and there is considerable variation in technique. Some studies omit comorbidity adjustment, some perform a simple comorbidity assessment or others a complex adjustment (presence and severity of comorbidity).
3. Time since treatment. For cross-sectional analysis, time since treatment must be controlled. For longitudinal, the follow-up times must be defined.
4. Response rate and missing data. While important for all QOL studies, for comparative treatment modality studies, additional selection bias may be entered through differences in response rates or frequency of incomplete responses. In general, a response rate of over 70% is desired and the treatment of missing data should be included in the methods, as this often leads to the exclusion of additional patients from the final analysis. For unknown reasons, minority groups have lower response rates.34'63 Telephone calls to remind patients to consider participating in the study are common and may improve response rates, but telephone interviews to obtain data for part or all of a study is controversial, as patients may respond differently to an interviewer and have questions explained that otherwise would not occur in interpreting the written questions alone.46
The PCOS trial has published a comparative study of RP versus XRT with 24 months of follow-up.63 SF-36 scores were mostly equivalent, although XRT scored lower for general health. They confirmed other studies demonstrating greater decreases in sexual and urinary function after RP, but more bowel side effect from XRT. Urinary bother mirrored loss of urinary function; however, sexual bother differed more by age than by sexual function - older patients reporting less bother.
The CaPSURE longitudinal study of RP versus XRT study also reports the same patterns of change, with XRT reporting declining sexual function from 1 to 2 years, with RP improving, such that the two groups were roughly equal by 2 years.64 Urinary function improved after the initial loss for RP, but remains lower than XRT at 2 years, and the bowel side effects of XRT improve to near RP scores by 2 years.
Madalinska etal.65 also performed a longitudinal study in the context of their clinical trial of prostate cancer screening in the Netherlands. They compared RP versus XRT using the SF-36 and UCLA PCI, also comparing patients who were diagnosed from screening clinics versus clinically detected. In contrast to the PCOS and CaPSURE studies that suffered from declining participation over time, the Madalinska study achieved a 91% baseline response rate, which held at 87% at 12 months. Despite theoretic selection biases, the baseline assessment of urinary and sexual function was not different between the groups. The same pattern of sexual, urinary and bowel function were noted, and the paper presented data to support the novel concept that screen-detected versus clinically detected cancers had the same quality of life.
Studies that include brachytherapy include Krupski et al.61 who compared RP versus brachytherapy with a cross-sectional analysis of the first year after treatment. Using the FACT-G, AUA and a separate urinary/sexual inventory, they found that RP and brachytherapy had the same FACT-G scores. Sexual and urinary scores were not significantly different. However, patients treated by brachytherapy/XRT showed more urinary symptoms and declining sexual function.
Lee etal.,62 as mentioned, have performed one of the few longitudinal studies that include brachytherapy, RP and XRT. As stated, using the FACT-G and IPSS surveys, all significant QOL changes were detected in the first 3 months of treatment and most had recovered to baseline by 12 months.
In the cross-sectional study by Davis etal.34 with 5 years of follow-up, brachytherapy patients had no incontinence or bowel dysfunction compared to age-matched norms. Urinary irritation was demonstrated with elevated AUA scores, which were mostly seen in patients treated within a year of the survey.
Of interest, urinary bother scores were not significantly different between RP and brachytherapy, thus supporting the notion that incontinence is not the only side effect of treatment important to patients. In terms of sexual function, brachy-therapy suffered sexual dysfunction that was more severe than XRT, but less severe than RP. However, both nerve-sparing RP patients and brachytherapy patients using erectile aids showed scores approaching or equal to age-matched controls.
Since validating the EPIC survey, the Michigan group66 has published their comprehensive cross-sectional analysis of brachytherapy, XRT and RP compared to age-matched controls. They included the SF-36, FACT-G and P, AUASI, and EPIC, and achieved a 72-79% response rate - a remarkable achievement for such a large set of questions. As with the previously cited studies, no significant differences were seen in the general health measures. However, with the EPIC survey's additional urinary obstruction/irritation, and hormonal domain, brachytherapy appeared just as morbid as RP and XRT. As with other studies, RP was associated with worse urinary incontinence and sexual function, and XRT with worse bowel and sexual function. By contrast, brachytherapy was associated with urinary irritation and obstruction, worse bowel function, and worse sexual function compared to controls. In the bother domains, brachytherapy reported significant bother for urinary, sexual and bowel; XRT reported significant sexual and bowel bother; and RP reported only sexual bother. Furthermore, in an analysis of long-term morbidity (>1 year after treatment), no significant difference was seen between RP and brachytherapy for urinary incontinence. In the hormonal domains, XRT and brachytherapy patients treated with neoadjuvant/adjuvant androgen deprivation reported lower function and bother scores. A unique finding of this study was that patients who had experienced a PSA recurrence reported significantly lower sexual and hormonal scores, and marginally lower general scores. Overall, in the opinion of these authors, brachy-therapy was less favorable than RP or XRT in several domains, and showed no advantage in any domain studied over a year after treatment.
As mentioned, this study was retrospective, single-institutional and cross-sectional, and may be biased by these factors. The next step, of course, will be longitudinal studies with pretreat-ment data. In the more distant future, the SPIRIT trial will study RP versus brachytherapy using the EPIC survey.
Was this article helpful?