The ProCaSP Study

Quality of Life Outcomes of Prostate Cancer Patients After Radiotherapy or Radical Prostatectomy in a Cohort Study

Nora Eisemann; Sandra Nolte; Maike Schnoor; Alexander Katalinic; Volker Rohde; Annika Waldmann


BMC Urol. 2015;15(28) 

In This Article


This study analysed changes of HRQOL of prostate cancer patients over time, the effect of treatment on HRQOL, and compared HRQOL to that in a German reference population.

Changes Over Time

The descriptive comparison of time trends showed that a decreased sexual functioning and limited recovery is more common in RP than in RT. Although the comparison does not convey information about a treatment effect on similar/randomised groups, it describes what happens to men with prostate cancer in actual health care. The finding is in concordance with other publications.[31,32] Further, better outcomes in erectile functioning for nsRP patients compared to patients with non-nerve-sparing procedures were also found in previous studies.[10,11,33] Still, erectile functioning is a concern in RP in general.[34–36]

In view of generic HRQOL, only minor changes over time were seen across groups. In several other studies these results have been ascribed to the fact that generic instruments are not able to a) distinguish adequately between highly selected groups[26] and b) adequately measure HRQOL after diagnosis because of response shift bias.[36,37] In this analysis, the application of multiple imputation may have obscured differences; the incorporation of uncertainty due to missing values often results in conservative estimates.

A clinically relevant worsening in PORPUS-P and an increase in emotional functioning at all measurement time points after baseline were observed in the RP group (with exceptions for emotional functioning in the nnsRP group). Differences between RP patients and norm values were close to clinical relevance at baseline (−8.7 and −9.3, respectively).[29] At the end of the follow-up differences in emotional functioning had largely disappeared, which may be ascribed to adaptation processes.[38] Our findings are in contrast to those of a Canadian study observing clinically relevant decreases of prostate-specific HRQOL after radiotherapy.[39] This is particularly interesting as over 50% of our RT patients, where information about pre-baseline androgen deprivation therapy was available, received ADT, i.e. an adjuvant therapy that negatively impacts erectile functioning, social functioning, and global health/HRQOL.[39] Hence, a clinically relevant decrease of HRQOL scores in our RT patients would have been expected. In a joint analysis of the Canadian data and our data, however, we were able to show that RT patients with ADT indeed scored lower HRQOL than RT patients without ADT.[40]

Treatment Effects

When controlling for baseline HRQOL, sexual functioning, and other possible confounders in a GEE model, the treatment effect of nsRP compared to nnsRP was favourable for many HRQOL domains and often unfavourable for RT. The treatment effect was significant for the domain diarrhoea of the HRQOL and for the PORPUS-P. The negative effect of RT on bowel functioning is in agreement with previous research.[41,42] However, treatment effects in our analysis never reached clinical relevance.

The treatment groups differed with respect to their baseline HRQOL, sexual functioning, and sociodemographic and tumour-related characteristics. Baseline HRQOL and sexual functioning was generally highest in patients receiving nsRP and lowest in patients receiving RT, especially in the combRT group. RT patients were older than RP patients, and the combRT group had the highest risk profile. Similar results have been reported in comparable international outcome studies[12,41] which is concordant with recommendations in current therapy guidelines such as the EAU-guideline advising an estimated 10-year survival as a precondition for RP but not for RT.[43] Logistic regression indicated that the significant predictors for a successful nerve-sparing surgery (in contrast to a non-nerve-sparing surgery) were a younger age, a higher Gleason score, and a better sexual functioning of the patient, while predictors for choosing radiotherapy were a better T-category, a worse sexual functioning, and an urban living area. However, there was considerable overlap with regard to these variables between the treatment groups, and the covariate adjustment in the regression analyses allowed to derive meaningful treatment effect estimates despite the group differences.

Comparison to Reference Population

Patients treated with RP had clinically relevant lower scores of fatigue and pain and higher role functioning than age-matched German men of the general population. The RT group also had lower fatigue and pain scores than suggested by the reference data but the difference was not clinically relevant (except for the externRT subgroup at baseline and the combRT subgroup after 24 months for pain). It can be suspected that the high proportion of participants reporting depression in the reference population caused artificially high norm values for fatigue and pain and low values for role functioning.[29] Even the 'low' symptom values (pain, fatigue) of the cancer patients in our study were higher than those reported by Krahn et al. (2009) for Canadian early stage prostate cancer patients.[39] The same applies to the role functioning values in the RP group.

Strengths and Limitations

Important strengths of the ProCaSP Study are the multicenter design, the comparison of HRQOL outcomes in several prostate cancer treatment groups, and the application of suitable statistical methods such as GEEs and adjustments for baseline differences. The main limitation of the study is missing data due to drop-out or incomplete questionnaires, especially for IIEF questions; however, by applying multiple imputation methods we achieved a sample size sufficient for stratification of two RP and three RT groups. The exclusion of urinary measures because of insufficient psychometric properties of the Prostate Specific Module limits the scope of our study. Finally, selection bias caused by an overrepresentation of severe cases in university hospitals cannot be ruled out. Possible confounding was handled by adjustment of covariates in the GEE model, although residual confounding cannot be excluded.