Previously presented in abstract form at the Annual Meeting of the American Society of Clinical Oncology, Atlanta, Georgia, June 2–6, 2006, and at the American Society for Therapeutic Radiology and Oncology Outcomes Research In Oncology Symposium, San Diego, California, September 15–16, 2006.
Pretreatment urinary, bowel, and sexual dysfunction may increase the toxicity of prostate cancer treatments or preclude potential benefits. Using patient-reported baseline dysfunction from a prospective cohort study, we determined the proportion of patients receiving relatively contraindicated (‘mismatched’) treatments.
Baseline obstructive uropathy and bowel dysfunction relatively contraindicate brachytherapy (BT) and external beam radiation therapy (EBRT), respectively, because they increase patients' vulnerability to treatment-related toxicity. Baseline sexual dysfunction renders moot the intended benefit of nerve-sparing radical prostatectomy (NSRP), which is to preserve sexual function. We categorized patients' clinical circumstances by increasing complexity and counted the mismatches in each, expecting weaker or multiple contraindications to increase mismatched treatments.
Of 438 eligible patients, 389 (89%) reported preexisting dysfunction, and more than one-third received mismatched treatments. Mismatches did not significantly increase with clinical complexity, and watchful waiting was very infrequent, even when all treatment options were contraindicated. Patient age and comorbidity, but not preexisting dysfunction, were associated with treatment choice. As expected, mismatched BT and EBRT led to worsened urinary and bowel symptoms, respectively, and NSRP did not improve outcomes after baseline sexual dysfunction.
Patients with early prostate cancer face an unusually broad range of treatment choices for which side effects, not efficacy, are often decisive. The 3 most common treatment modalities, external beam radiation (EBRT), brachytherapy (BT), and radical prostatectomy (RP), have not been shown in controlled trials to differ in efficacy, yet each causes a distinctive pattern of bowel, urinary, and sexual dysfunction.1–9 Along with treatment modality, preexisting patient dysfunction is the most important predictor of quality of life outcomes.1, 6 Because such dysfunction may indicate a patient's increased vulnerability to a specific treatment modality or his inability to benefit from it, pretreatment dysfunction needs to be explicitly communicated between patient and physician and factored into the treatment decision-making process. In a prospective cohort study of early prostate cancer treatment outcomes, we used validated survey instruments to assess bowel, urinary, and sexual dysfunction before treatment and at fixed intervals afterwards. The results provide an opportunity to use patient-reported pretreatment dysfunction to determine whether their treatment choices incorporated this information.
We defined treatment ‘mismatches’ as instances when patients received primary therapies that would 1) be especially likely to worsen baseline dysfunction, or 2) be unlikely to achieve the intended functional benefit. For example, patients who have baseline bowel dysfunction would be ‘mismatched’ if their primary therapy were EBRT, a treatment modality that inevitably irradiates the adjacent rectum in addition to the prostate, causing acute and long-term bowel dysfunction.1, 2, 5 The alternative treatments, RP and BT, are less likely to worsen the patients' preexisting bowel dysfunction.1, 5 Similarly, patients with baseline urinary obstructive/irritative symptoms are ‘mismatched’ with BT, which delivers a high dose of radiation in the prostate, increasing the likelihood of painful complete obstruction in patients with preexisting obstructive symptoms.10, 11 Therefore, RP and EBRT are the preferred treatments. Alternatively, patients with baseline erectile dysfunction would not likely benefit from nerve-sparing radical prostatectomy (NSRP), a technique that in attempting to better preserve sexual function prolongs surgery and may increase the likelihood of incomplete excision of potentially cancer-containing prostatic tissue.12 Because baseline erectile dysfunction markedly reduces or eliminates its intended quality of life benefit, whereas the disadvantages remain, we defined NSRP in such patients as mismatches; nonnerve-sparing radical prostatectomy (NNSRP) would produce a similar functional outcome with less potential harm. Surgical patients with good sexual function at baseline who underwent NNSRP may not have had tumors amenable to NSRP, and the patients may not have preferred a procedure with potential disadvantages regarding surgical excision or surgical duration. Therefore, we did not consider NNSRP to be a ‘mismatched’ treatment for these patients.
To capture some of the complexity of medical decision-making, which involves competing tradeoffs, patient preferences and physician practice patterns, we classified patients into 4 groups of increasing clinical complexity (Table 1), using the level of preexisting dysfunction (‘intermediate’ vs ‘poor’)9 and the number of mismatched modalities. We then examined the proportion of patients receiving mismatched treatments in each group, expecting both lesser preexisting dysfunction and increasing mismatched modalities to increase the frequency of mismatches. For example, a patient with severe bowel dysfunction but no other relevant dysfunction at baseline has the most straightforward and forceful contraindication against EBRT, the single treatment most likely to worsen the patient's condition. Because we expect treating physicians aware of the patient's status to strongly recommend alternatives to the contraindicated treatment, even in the face of contrary patient preferences, we expected treatment mismatches to be uncommon in this group of patients. However, other circumstances are less clear-cut. A lesser dysfunction at baseline weakens the contraindication, whereas dysfunction in multiple organ systems narrows the treatment choices, perhaps forcing reconsideration of otherwise contraindicated treatments.
Table 1. Definition of Patient Groups
Severe baseline dysfunction in one organ domain, relatively contraindicating one treatment option.
Most straightforward and forceful contraindication against one treatment option.
Moderate baseline dysfunction in one organ domain, relatively contraindicating one treatment option.
Moderate dysfunction weakens the contraindication, but multiple “appropriate” options remain.
Baseline dysfunction in multiple organ domains, relatively contraindicating multiple treatment options. At least one “appropriate” option available.
Patients are more likely to reconsider contraindicated treatments when choices are significantly limited.
All active treatments relatively contraindicated.
Watchful waiting may be more common in these patients.
Each of these changes in circumstances, less severe baseline dysfunction that attenuates the contraindication and multiple contraindications that reduce the available options, increases decision-making ambiguity compared with a single stronger contraindication. The result allows greater sway for patient and physician preferences in the treatment-decision process, potentially increasing the likelihood of treatment mismatches. We also examined patient outcomes to assess our assumption that mismatched treatments produced worse outcomes or intended benefits that remained unrealized.
MATERIALS AND METHODS
Patients of participating investigators with pathologically confirmed, untreated, clinically localized prostate cancer seeking consultation at Massachusetts General Hospital, Dana-Farber Cancer Institute, Beth Israel-Deaconess Hospital, or MetroWest Medical Center between June 1, 1994, and August 31, 2000, were eligible, whether or not they received treatment elsewhere. Most patients were recruited from multispecialty clinics, representing a small subset of patients with the diagnosis of prostate cancer at these institutions who we felt had received more balanced and comprehensive clinical advice than those who consulted with only a single specialty.
Institutional Review Boards at all participating institutions approved the protocol. In general, study patients were socioeconomically advantaged, with 98% Caucasians, the median education level a college degree, and the median income over $50,000.
Data Collection and Instruments
We described the study instruments, instrument validation, and data collection procedures in detail previously.1, 13 Briefly, at the initial clinic consultation patients were given an invitation letter describing the study, along with the baseline questionnaire, which was completed before initiating primary treatment and mailed by the patient directly to us. Subsequent questionnaires were mailed to enrolled patients at 3, 12, 24, and 36 months after treatment began. We obtained patient permission to review their medical records to determine cancer stage, preexisting comorbidities, measured by the Index of Coexistent Disease (ICED),14 and initial treatment received.
The baseline questionnaire included demographic items such as date of birth, ethnicity, highest education level, employment, and marital status, and all questionnaires included 4 symptom indices to assess patient-reported urinary incontinence, urinary obstruction/irritation, bowel dysfunction, and sexual dysfunction. The Urinary Incontinence Index contains 3 questions gauging the degree of urinary control. The Urinary Obstruction and Irritation Index contains 5 questions assessing hesitancy, frequency, nocturia, dysuria, and urgency. The 5 Sexual Dysfunction items assessed patient erections (firmness, difficulty getting and keeping erections), ejaculation, and orgasm. Bowel problems items include diarrhea, bowel urgency, pain with bowel movements, rectal bleeding, abdominal cramping, and tenesmus. For each index we asked parallel questions to assess the level of distress patients felt from each symptom.
Each index was scored by summing the component items and then standardizing that value to vary from 0 (no dysfunction) to 100 (maximum dysfunction). However, the clinical meaning of numerical index values can be elusive. Therefore, we categorized symptoms scores into ‘normal,’ ‘intermediate,’ and ‘poor’ levels by examining the corresponding distress indices. This methodology was described in detail elsewhere.9 ‘Normal’ function indicates the range of symptom scores corresponding to minimal patient distress, whereas ‘intermediate’ and ‘poor’ function correspond to moderate or severe distress, respectively. In this way, index scores were translated to clinically meaningful, psychometrically coherent, ordered clusters of symptoms with minimal loss of the detailed, patient-reported information used in the validated multi-item scale.
We classified patients by their initial primary treatment (RP, EBRT, or BT). At least 2 urologists independently determined the surgical technique (NSRP or nonnerve-sparing RP, NNSRP) of each RP patient, based on operative reports.
Age and symptom indices were considered continuous variables. All other variables were considered ordinal or dichotomous measures. We compared treatment groups using the Kruskal-Wallis test for continuous and ordinal variables. Fisher exact test and chi-square for trend were used for dichotomous variables. Outcomes data are reported as proportions for clarity, but we used the Wilcoxon Signed Rank test in comparing the same group of patients at different timepoints. SAS v 9.1 (Cary, NC) was used for all data analysis and all reported P-values are 2-sided.
Baseline Patient Characteristics
Six hundred thirteen patients were eligible. Of these, 91 patients dropped out before the first follow-up questionnaire at 3 months, leaving 522 participating patients. An additional 84 patients (16%) dropped out before completing the 36-month questionnaire, leaving 438 patients (84% of eligible participating patients) followed through 36 months.
The primary treatment for these 438 patients was RP for 127 patients (29%), of whom 74 received nerve-sparing surgery; EBRT for 190 patients (43%), of whom 63 received concurrent androgen-deprivation therapy; BT for 92 patients (21%); watchful waiting for 19 patients (4%) (Table 2); and other treatments (hormonal ablation alone and cryotherapy) for 10 patients (2%). Patients who received hormonal ablation alone and cryotherapy were too few for meaningful analysis and were excluded. As expected and as we reported earlier, RP (especially NSRP) and BT patients had more favorable cancer characteristics compared with EBRT patients at baseline.1 Further, they were younger (P < .001) and had lower ICED scores (P < .01) than those who chose EBRT or watchful waiting.
At baseline, 389 patients (89%) reported intermediate or poor function relevant to treatment, including 266 patients (61%) with urinary obstruction/irritation (164 intermediate, 102 poor), 281 patients (64%) with sexual dysfunction (115 intermediate, 166 poor), and 215 patients (49%) with bowel dysfunction (177 intermediate, 38 poor). Overall, 120 patients (27%) reported only 1 dysfunction, 165 patients (38%) reported 2, and 104 patients (24%) had dysfunction in all 3 relevant symptom domains at baseline.
Mismatched Treatments Received
Clinical decision-making involves competing priorities, patient preferences and physician practice patterns. To better represent actual clinical decision-making, we grouped patients into scenarios of increasing clinical complexity, either because of less severe pretreatment dysfunction or multiple mismatched treatment modalities, and recorded the frequency of mismatched treatments by group (Table 3). Group 1, containing 50 patients, represents the least ambiguous scenarios with the strongest and most straightforward contraindications. These patients have a single, severe (poor) baseline dysfunction that forcefully contraindicates only 1 active treatment: bowel problems that contraindicate EBRT, urinary obstruction/irritation that contraindicate BT, or with sexual dysfunction obviates the benefit of NSRP. We expected the fewest treatment mismatches in this group. Group 2 includes patients with less severe (intermediate) baseline dysfunction that contraindicates a single treatment with less strength, allowing greater sway for patient and physician preferences in choosing treatment. We expected more mismatches in this group, given a weaker contraindication for competing considerations to overcome. Group 3 includes patients with dysfunction in multiple relevant domains at baseline, contraindicating multiple treatment choices, but with at least 1 remaining ‘appropriate’ (not contraindicated) option available. Given more constrained treatment options and competing contraindications, we expected still more mismatches in this group. Group 4 includes patients with contraindications to all 3 active treatment options. Here, we introduce older age, which we define as 65 years or more, as a widely observed relative contraindication to surgery due to increased surgical risk, during the time of this cohort study. Thus, no active treatment option would remain for patients older than 65 for whom EBRT and BT were also contraindicated. Given the apparent contraindication to each active treatment, we expected watchful waiting to be more common among these men than those in Groups 1 to 3.
Table 3. Frequency of Mismatched Treatments as a Function of Clinical Complexity
Of Group 1 patients, with a single strong contraindication, 34% received mismatched treatments (Table 3). Mismatches did not rise significantly with increasing clinical complexity, to 37% in Group 2 and 40% in Group 3 (chi-square for trend, P = .71). Further, of Group 4 patients, for whom all treatment options were contraindicated, only 5% chose watchful waiting, no different from the remaining groups (Fisher exact, P = .63).
To explore other factors that might motivate patients to override contraindications and produce mismatches, we examined factors that might make preserving sexual function more appealing, regardless of how unlikely. We found patients with baseline sexual dysfunction who underwent NSRP were no more likely to be married (85% for NSRP vs 91%, Fisher exact, P = .69) or have a regular sexual partner (72% vs 85%, Fisher exact, P = .33) compared with patients who underwent NNSRP. However, NSRP patients were, on average, younger than NNSRP patients (58.2 years vs 61.4, P < .01), and tended toward less baseline medical comorbidity (ICED score, P = .10).
Given that published information and clinical emphasis on treatment-related quality of life in early prostate cancer has increased over the last decade, we examined the rate of mismatches over time by dividing our cohort into thirds based on date of study entry. The rate of mismatches did not decrease over time (chi-square for trend, P = .52).
Not surprisingly, patients with preexisting urinary or bowel dysfunction had worsened function after mismatched treatments. Patients with baseline urinary obstruction/irritation who received BT more often reported nocturia (78% pretreatment to 90% posttreatment, P = .06) and dysuria (8%–34%, P < .01). Also, preexisting symptoms at baseline, including nocturia, dysuria, urinary frequency, and urgency, worsened (data not shown). More patients with baseline bowel dysfunction who received EBRT reported diarrhea (32% pretreatment to 43% posttreatment, P = .02), pain with bowel movements (7%–33%, P < .001), bowel urgency (32%–43%, P < .001), and rectal bleeding (8%–32%, P < .001). Patients with baseline bowel dysfunction but no rectal bleeding tended to develop it after EBRT more often than those with normal baseline bowel function (28% vs 17%, P = .08). Preexisting diarrhea, pain with bowel movements, rectal bleeding, and bowel urgency became more severe after EBRT.
Patients who had baseline sexual dysfunction did not derive additional benefit from having nerve-sparing RP, compared with NNSRP. All 30 patients with intermediate sexual function at baseline reported at least occasional erections adequate for sexual intercourse before surgery, but sexual function for the 15 NSRP and 15 NNSRP patients declined in parallel after surgery (Fig. 1). At 36 months, 40% of NSRP patients and 36% of NNSRP patients reported erections at least occasionally adequate for sexual intercourse (P = .85). Only 1 patient in each group with intermediate pretreatment sexual function had preserved it long-term, whereas the rest had declined to poor function. Similarly, in patients with poor sexual function at baseline, NSRP and NNSRP produced equally poor results. No patient who underwent RP with poor sexual function improved.
The association between NSRP and better preserved urinary continence may be another possible reason for choosing it, although evidence on this is mixed.15–19 Among patients with baseline sexual dysfunction, urinary incontinence increased for both NSRP patients (mean score 6.3 at baseline to 17.9 at 36 months, change 11.6) and NNSRP patients (5.2 to 24.2, change 19.1) (P = .36).
Quality of life considerations play an important role in treatment decisions for early prostate cancer, as there is little reliable evidence that the 3 major treatment modalities, EBRT, BT, and RP, differ in efficacy, whereas their distinctive patterns of treatment-related organ dysfunction are well established.1–9 Preexisting urinary, bowel, and sexual dysfunction may make patients particularly vulnerable to or unable to benefit from 1 or more treatment modalities. Avoiding such treatment ‘mismatch’ requires factoring relevant pretreatment dysfunction into the decision-making process. In this study, we compared patient-reported pretreatment dysfunction to their subsequent treatments as an indicator of physician-patient communication on these sometimes uncomfortable but essential topics. The mismatched treatments we identified using questionnaires outside the clinical setting may be a useful indicator that the information was not incorporated into treatment decisions as it should have been. The results provide reason for concern. We expected mismatches to be infrequent, especially when the contraindications are strong and confined to 1 of several available treatment modalities, but to increase significantly as the clinical scenario became more complex; that is, when the contraindicating baseline dysfunction was less forceful (intermediate vs poor) or when dysfunction involved multiple organ systems contraindicating other treatments, further reducing options.
We were surprised by the proportion of our relatively young, healthy, and prosperous population with relevant baseline organ dysfunction (89%), and the proportion of such patients who received mismatched treatments. A third of patients in Group 1, each of whom had a single severe baseline dysfunction that contraindicated 1 treatment, received the mismatched treatment. Increasing clinical complexity had little apparent effect on the likelihood of mismatched treatments. The chance that a patient in Group 1 underwent a mismatched treatment did not differ from that of a Group 2 patient, whose contraindication was weaker, or a Group 3 patient, for whom multiple treatments were contraindicated. These results provide surprisingly little evidence that relevant pretreatment organ dysfunction influenced the treatment decisions.
Understanding that clinical decision-making is complex, requiring balancing competing factors, we explored possible causes for the high rate of treatment mismatches. In contrast to pretreatment dysfunction, patient age, and medical condition, which are routinely documented and do not require sensitive communication, clearly influenced the choice of active treatment: younger patients with less baseline medical comorbidity more often underwent RP than EBRT and watchful waiting, consistent with physician preferences. Further, consistent with widespread practice patterns, very few patients in our cohort chose watchful waiting. Even among Group 4 patients, who had no available ‘appropriate’ treatment choice at all, only 5% chose watchful waiting. Instead, Group 4 patients usually underwent EBRT, reflecting both the physicians' and patients' joint bias toward active treatment, and physician preference for EBRT for older patients with greater baseline comorbidity. Of plausible causes of inappropriate NSRP in impotent men, only younger age appeared influential, although less comorbidity (lower ICED scores) approached a significant relation. Mismatched treatments did not decline over time, as treatment-related quality of life has become more widely appreciated and clinically emphasized.
Whereas we found little impact of pretreatment dysfunction on treatment decision-making, we cannot rule out other explanations for our results. Although we are confident we can measure dysfunction,1 and that our specific contraindications are clinically sensible, we are aware that patient and physician preferences or other factors may overrule them in choosing treatment. Further study is required to determine whether pretreatment dysfunctions were not identified, not addressed, or appropriately overridden. Further, the patient decision-making process may in fact not conform to rational assumptions. Denberg et al.20 found that 20 prostate cancer patients recounted hurried decisions, based on unchallenged misconceptions and anecdotes. However, we hope that physician recommendations, the most powerful influence on patient treatment decisions, would reflect more rational processes.
However, some known factors may have impeded communication. Although patient age and comorbidities are routinely available in patient-physician encounters, quality of life information requires specific, sometimes uncomfortable questions and candid responses. Several prostate cancer-specific studies have found much greater dysfunction when patients, not physicians, report it,2, 21–23 and randomized trials found, in clinical settings, patient survey results increased physician awareness of patient symptoms over the standard interview.24, 25 Patients report different things to their doctors than in questionnaires. Salonia et al.23 found that of 234 prostate cancer patients who verbally reported full sexual potency and were scheduled to undergo NSRP, 57% reported baseline erectile dysfunction using the International Index of Erectile Dysfunction questionnaire, including 25 patients who did not even attempt intercourse during the 4 weeks before surgery.23 Routine use of preconsultation questionnaires may bring to light dysfunction that may reduce treatment mismatch.
Patient results confirmed the postulated mismatches, as such treatments produced worse outcomes. BT produced increased symptoms in patients with baseline urinary obstruction/irritation, and EBRT increased diarrhea, painful bowel movements, bowel urgency, and rectal bleeding in patients with baseline bowel dysfunction. We found no benefit from NSRP in patients with significantly compromised sexual function before surgery, in either sexual function or urinary incontinence, another benefit sometimes put forward to justify the procedure.
The competing treatment choices early prostate cancer patients face, distinguished by quality of life but not efficacy outcomes, provided an opportunity to explore the impact of pretreatment dysfunction on treatment decision-making as a potential indicator of physician-patient communication. Although we cannot be certain that pretreatment dysfunction was not fully considered, the frequency of mismatched treatments and the limited impact of increasing clinical ambiguity is difficult to reconcile with the assumption that it is. If so, incorporating validated prostate cancer-specific survey instruments into clinical practice may increase awareness of patient baseline dysfunction, reduce treatment mismatches, and improve outcomes. We will evaluate this hypothesis prospectively. Further, these results may have more general significance in oncology. If organ-specific dysfunction is poorly communicated in prostate cancer, where its importance in treatment decision-making is well established, it may be even more prevalent in other malignancies, producing unsuspected barriers to patient-centered choices of treatment, palliative care and hospice.
This study has several limitations. Small sample sizes limited the statistical power of some comparisons. As we have noted, the process of patient treatment decision-making is complicated. Whereas we identified treatments relatively contraindicated by patient urinary, bowel, and sexual dysfunction and age, other options may have been constrained by factors we did not identify, and patient preferences may have overridden these relative contraindications. However, our observations raise concerns about physician-patient communication, suggest a correctable threat to high-quality patient-centered care, an outcome to measure it, and an intervention to address it.