MORE THAN 140,000 men in the United States annually

Time Course and Predictors of Symptoms After Primary Prostate Cancer Therapy By James A. Talcott, Judith Manola, Jack A. Clark, Irving Kaplan, Clair J. Beard, Sonya P. Mitchell, Ronald C. Chen, Michael P. O Leary, Philip W. Kantoff, and Anthony V. D Amico MORE THAN 140,000 men in the United States annually are diagnosed with early (clinically localized) prostate cancer (PC). They must choose one of three local treatments, the long-term side effects of which differ, or choose initial observation, avoiding treatment complications but forgoing a chance to prevent eventual metastatic cancer. However, no local treatment has been proven to prolong life; most patients die of unrelated causes, and most PC deaths occur more than a decade after diagnosis. Most patients undergo active therapy with external beam radiation therapy (EBRT), radical prostatectomy (RP), interstitial radioactive seed implants, or brachytherapy (BT), each of which has a distinctive pattern of long-lasting urinary, bowel, and sexual dysfunction. 1-7 Weighing potential treatments, particularly without a documented survival benefit, requires accurate information on quality-of-life consequences for each treatment. Unfortunately, studies assessing treatment complications are methodologically limited and, because none are randomized, biased. Patients choose and are advised to have specific treatments because of personal and clinical characteristics that in turn influence their outcomes. These factors include age, PC prognostic factors, comorbid medical conditions, and pretreatment function of systems potentially damaged by treatment, such as impotence or bowel problems. Therefore, to assess the consequences of treatment, potential confounding factors must be measured and, for factors subject to recall bias, assessment must occur before treatment. 8 Data collection technique strongly influences results. Physicians discover and report far fewer complications than patients do, 1,9,10 and varying complication Purpose: Understanding the distinctive patterns of treatment-related dysfunction after alternative initial treatments for early prostate cancer (PC) may improve patients choice of treatment and later help them adjust to its consequences. We characterized the time course of treatment complications while adjusting for potentially confounding pretreatment factors hindering other observational studies. Patients and Methods: In a prospective cohort study of 417 men we assessed urinary, bowel, and sexual function from before primary treatment to 24 months after. To control for potential confounding, we measured sociodemographic and PC prognostic factors, medical comorbidity, and pretreatment function commonly affected by PC and its treatment. Results: Patients who underwent external beam radiotherapy (EBRT), radical prostatectomy (RP), and brachytherapy (BT) differed significantly in sociodemographic factors, cancer prognostic factors, and pretreatment symptom status, especially sexual function. Urinary incontinence increased sharply after RP, while bowel problems and urinary irritation/obstruction rose after EBRT and BT. Sexual dysfunction increased in all patients, particularly after radical prostatectomy, and nerve-sparing surgical technique had little apparent benefit. There was no change in urinary function and little change in overall bowel function after 12 months, but the time course of sexual dysfunction varied by treatment and, for bowel function, by symptom. Multiple regression modeling confirmed that treatment influences all 24-month outcomes, but residual confounding persisted. Conclusion: Pretreatment function and the primary treatment modality for early stage PC strongly predict the affected organ systems and time course of dysfunction. With this information, patients and their physicians may refine their choice of treatment and better anticipate its consequences. J Clin Oncol 21:3979-3986. 2003 by American Society of Clinical Oncology. definitions produce widely varying rates. 11 Systematic use of multiple reports, such as those of Wasson et al 12 and Robinson et al, 13 produce more precise and generalizable estimates, but preserve the biases of the methodologically limited studies they include. Finally, knowing the time course of each treatment s complications may help patients anticipate and adjust to their chosen treatment s effects. In our previous cohort study of men with untreated early PC, 10,14,15 we found sharply increased bowel symptoms after EBRT, and urinary incontinence and sexual dysfunction after RP at 3 months after treatment, but modest improvement by 12 months. However, the study s 1-year follow-up was short, its outcomes measures were incomplete and imprecise, and it From the Center for Outcomes Research, MGH Cancer Center, Massachusetts General Hospital; Boston University School of Public Health; Dana-Farber Cancer Institute; Brigham and Women s Hospital; Beth Israel-Deaconess Medical Center; Harvard Medical School, Boston; Center for Health Quality, Outcomes, and Economic Research, Edith Nourse Rogers Memorial Veterans Hospital, Bedford, MA. Submitted January 27, 2003; accepted August 7, 2003. Supported by an Agency for Healthcare Research and Quality grant, HS08208. Authors disclosures of potential conflicts of interest are found at the end of this article. Address reprint requests to James A. Talcott, MD, Center for Outcomes Research, Massachusetts General Hospital, B75 230, 55 Fruit St, Boston, MA 02114-2696; e-mail: jtalcott@partners.org. 2003 by American Society of Clinical Oncology. 0732-183X/03/2121-3979/$20.00 Journal of Clinical Oncology, Vol 21, No 21 (November 1), 2003: pp 3979-3986 DOI: 10.1200/JCO.2003.01.199 3979

3980 TALCOTT ET AL included few BT patients. To address these issues, we began a study of 600 newly diagnosed patients with longer followup, an improved, validated quality of life instrument, and more BT patients. To better define the time course of treatment-specific dysfunction and the factors influencing it, we now report on 417 patients in the expanded cohort study with at least 24-month follow-up. Patient Population PATIENTS AND METHODS We recruited patients between June 1, 1994, and August 31, 2000, when they consulted physicians at participating sites for advice on primary therapy of pathologically-confirmed, untreated early (nonmetastatic) PC at the Massachusetts General Hospital, the Dana-Farber Cancer Institute, and affiliated clinics of the Joint Center for Radiation Therapy (all in Boston, MA). Because the use of BT increased rapidly, we extended enrollment for BT patients to include at least 100 such patients. Because we did not require treatment at a study site, all study enrollment occurred at teaching hospitals of Harvard Medical School (Boston, MA), but treatment could occur elsewhere. Institutional Review Boards at all participating institutions approved the protocol. Data Collection and Instruments Data collection. In a letter describing the study, we invited eligible patients to complete an accompanying baseline questionnaire before beginning primary therapy and at 3, 12, 24 and 36 months after treatment began, or after study enrollment if watchful waiting or initial observation was chosen. We also requested that patients identify all institutions and physicians who treated their PC and give permission to review their medical records. From medical records we recorded cancer stage, preexisting medical comorbidity using the Index of Coexistent Disease (ICED), which assigns a score of 0 to 4 based on both disease severity and functional limitations, 16 initial anticancer treatments received, and subsequent medical events. We sent follow-up questionnaires at the appropriate intervals, with a written and telephone reminder for nonreturned questionnaires. Study instruments. Patient-completed questionnaires addressed demographic descriptors, global health-related quality of life using the Medical Outcomes Study 36-Item Short-Form Health Survey (SF-36), 17 and treatment-related symptoms. Urinary, bowel, and sexual dysfunction were assessed by four symptom indexes and four parallel symptom distress scales. The details of their development and validation are described elsewhere. 18 In brief, we designed focused, structured patient-report items to comprehensively assess symptoms of urinary incontinence, urinary obstruction/irritation, bowel dysfunction, and sexual dysfunction. These patient-report items were based on previously published questionnaires, the results of our pilot study, and the clinical experience of genitourinary medical, radiation and surgical oncologists on our research team. The urinary incontinence index assessed the degree of urinary control and the frequency and magnitude of leakage in men who had less than complete control. Five items assessed urinary obstruction and irritation: hesitancy, frequency of urination during the day, nocturia (urination at night), dysuria, and urgency. The items for bowel problems included diarrhea, urgency of bowel movements, pain, bleeding, passing mucus during bowel movements, abdominal cramping, and tenesmus. Sexual function items focused on the firmness of erections, difficulty in getting and keeping erections, and frequency of ejaculation and orgasm. Four other indexes assessed distress attributed to these symptoms. However, because these were highly correlated with functional scales in our psychometric evaluation, we do not present them here. Scores ranged between 0 (minimum) and 100 (maximum dysfunction). Physical and mental component summaries of the SF-36 were calculated to characterize pretreatment functional health status. 19 Because SF-36 scales change only transiently in this setting, they were not used to assess outcomes. 20 Treatment received. We classified patients by their initial primary treatment (RP, EBRT or BT). Hormone ablation was considered an additional treatment, since it was usually combined with a local therapy. To determine the surgical technique patients received, at least two urologists blindly determined from each patient s operative report whether the surgeon had preserved none, one, or both neurovascular bundles; a third urologist was asked to resolve any conflict. Statistical Methods We compared treatment groups by baseline symptoms and sociodemographic characteristics using Fisher s exact test for categorical measures and Wilcoxon rank sum tests for continuous measures. Effect sizes (ES) were calculated by dividing mean score difference by the standard deviation (SD) at baseline. We used linear regression univariate and multivariable models to explore relationships between outcomes (symptom scores at 24 months for each of the four scales) and treatment and baseline covariates, including risk categories defined by D Amico et al. 21 Low-risk patients had prostatespecific antigen (PSA) less than 10, Gleason grade less than 7, and stage T1b or T2a disease. Intermediate-risk patients had PSA 10 to 20 ng/ml, Gleason grade 7, or stage T2b disease. High-risk patients had PSA greater than 20, Gleason grade 8 to 10, or tumor stage T2c or higher. Gleason category (4 to 6, 7, 8 to 10), risk category, and Index of Coexistent Disease score (0,1, 2 or more) were considered categorical variables with equal spacing between levels. Except for age, which was continuous, all remaining variables were dichotomous. We defined current marital status and education level (some graduate school or not) at study entry. We used a forward selection procedure to build multivariable models, using a cutoff P value of.10 for entry into the model, and using data only from patients who had nonmissing values for all candidate variables. Symptom scores at baseline and treatment were considered mandatory variables and were included in all models. The final models were rerun with all patients with data for the selected variables. Models were then reviewed to assure that they made sense from a clinical perspective. RESULTS Baseline Patient Characteristics Baseline questionnaires from 613 eligible patients were returned within 7 days of initiating treatment, a requirement to ensure that the reported baseline symptoms were free from the effects of therapy. As of December 31, 2000, 91 enrolled patients (14.9%) dropped out or died of unrelated causes (four patients) before the first follow-up questionnaire at 3 months after treatment, leaving 522 participating patients. Of the remaining patients, 27 patients (11%) had not yet been followed for 24 months, 66 patients (12.5%) dropped out before completing the 24-month questionnaires, and 429 patients (76.6%) continued through 24 months. Patients who dropped out did not differ from continuing patients in age, PC prognostic factors, pretreatment SF-36 summary scores, or urinary, bowel, or sexual function or bother scores, other than slightly increased obstruction/irritation. We could not obtain the medical records of 12 continuing patients. The primary therapy for the remaining 417 patients was EBRT for 182 patients (44%), of whom 65 also received androgen-deprivation therapy, 18 of them starting either a month or more before completing the questionnaire or at an unknown time, RP (including five procedures aborted after cancer was found in the preceding lymph node dissection) for 129 patients (31%), BT for 80 patients (19%), initial observation for 19 patients (5%), and other treatments (cryotherapy or hormone ablation alone) for seven patients (1%; Table 1). Treatment groups differed in initial characteristics, demonstrating the potential for confounding between treatment groups

SYMPTOMS AFTER PROSTATE CANCER TREATMENT 3981 Table 1. Initial Treatments for Early Prostate Cancer Received by 417 Patients With Known Treatment and at Least 24 Months Follow-Up Study Patients Treatment Group Patients Treatment Subgroup Patients Treatment* Observation (no initial 19 5 treatment) All EBRT 182 44 EBRT alone 117 64 EBRT hormonal therapy 65 36 All radical prostatectomy 129 31 Non nerve sparing 48 37 Nerve sparing 71 55 Unilateral 39 55 Bilateral 26 37 bundles unknown 6 8 Nerve sparing status unknown 10 8 Brachytherapy 80 19 Hormones alone 5 1 All cryotherapy 2 1 Total 417 100 Abbreviation: EBRT, external-beam radiotherapy. *Twelve patients with incomplete documentation of treatment were excluded from analysis. seen in other observational studies (Table 2). Study patients were socioeconomically advantaged, with 86% currently married. About half the population, which averaged 65 years old, was currently employed, with a median reported family income over $50,000. Nearly all were white. The median education level was a college degree, and a third of the population had a graduate or professional degree. Patient age varied by treatment; RP patients averaged 4 years younger than BT patients (P.001), who in turn were 5 years younger than EBRT patients (P.001) at enrollment. Other sociodemographic variables reflected the age differences; the older EBRT patients were less likely to be employed and had lower current income than RP and BT patients (P.002), and had less education than RP patients (P.003). Medical prognostic factors favored BT patients (who were increasingly carefully selected during this period) and RP patients. Both groups had less documented medical comorbidity (P.03) and lower PSA levels (P.01) than EBRT patients. BT patients less often had Gleason scores over 6 (P.0001) and intermediate or high risk designation by the criteria of D Amico et al (P.009). 21 Patient-Reported Symptoms Before treatment, urinary incontinence was infrequent, and did not differ significantly by treatment group (Table 3). The low baseline scores reflect low reported incontinence (15%) and only rare daily leaking (1%) in the prior week. After EBRT and BT incontinence did not change significantly, but rose sharply after RP from 4.9 to 35.0 at 3 months (ES, 2.5), with similar changes in bother scores (data not shown). Among patients receiving nerve-sparing surgery, the increase was less. By 12 months, RP patients mean score had improved to 23.9 (ES, 1.6) but did not improve thereafter. At 24 months, 57% of surgery patients reported incontinence within 7 days, 17% reported incontinence at least daily, and 20% reported wearing absorptive pads. Radiation treatment, especially BT, transiently increased urinary obstruction/irritation. BT patients mean dysfunction score rose from 16.0 to 25.1 (ES, 0.9) at 3 months, with a smaller rise for EBRT patients. Since obstruction symptoms are subjectively most intense shortly after brachytherapy, our 3-month assessment may have missed greater dysfunction occurring earlier. Reported bowel dysfunction was also minimal before treatment. However, at 3 months, shortly after completion of EBRT, bowel problems scores had risen from 4.6 to 12.2 (ES, 1.2). The proportion of men reporting at least occasional diarrhea had risen from 17% to 43%, and 13% reported diarrhea at least several times a week. Mean bowel problems scores improved at 12 and 24 months (although not to baseline) but the overall scores obscured disparate trends for the individual symptoms that comprise the index. Among EBRT patients, the frequency of diarrhea had returned to baselines by the 24 month follow-up point, tenesmus improved less, bowel urgency did not improve after 3 months, and the reported prevalence of rectal bleeding increased steadily from 5% at pretreatment to 25% at 24 months. Pretreatment sexual dysfunction was lowest among RP patients, although that of BT patients was almost as good (Table 3). However, sexual dysfunction had increased sharply (change from baseline to 3 months, 54.4 points; ES, 1.7) when first assessed at 3 months after RP. During this initial interval, EBRT and BT patients sexual function decreased modestly (ES, 0.3 to 0.4), while the observation group did not begin to show a decline until 12 months. Modest subsequent improvement by RP patients combined with the continuing increase in other groups dysfunction reduced the relative RP disadvantage at 24 months. Factors associated with sexual dysfunction at 24 months are presented in Table 4. Not surprisingly, patients who underwent nerve-sparing surgery had better pretreatment sexual function, and it remained

3982 TALCOTT ET AL Table 2. Sociodemographic and Clinical Characteristics of 410 Patients With Early Prostate Cancer Who Chose Surgery, Radiation, or Initial Observation and Completed 24-Month Participation Characteristic Radical Prostatectomy External Beam RT Brachytherapy Initial Observation Total patients 129 182 80 19 410 Age, years Median 60 69 64 67 65 Mean 60 69 63 68 65 Range 46-74 51-82 47-76 54-79 46-82 Race or ethnicity White 99 98 93 100 98 Black 1 1 0 0 1 Asian 0 0 3 0 1 Native American 0 1 0 0 1 Other 0 0 4 0 1 Unknown 1 4 3 8 Currently married 91 84 82 84 86 Unknown 1 3 2 6 Currently employed 61 39 68 32 51 Unknown 1 2 2 5 Current income $40,000 23 46 10 47 32 $40,001-$70,000 32 38 47 35 37 $70,000 46 16 44 18 31 Unknown 8 15 7 2 32 Highest educational level High school or less 9 23 13 11 16 Some college to some graduate school 48 46 50 53 48 Grad or professional degree 42 31 37 37 36 Unknown 1 3 2 6 Physical component summary of SF-36 Median 56.1 52.4 55.6 55.8 54.6 Mean 54.9 50.2 53.4 52.5 52.4 Range 27.4-68.2 21.3-68.4 27.5-63.2 31.4-69.5 21.3-69.5 Mental component summary of SF-36 Median 52.6 54.3 55.2 56.8 54.1 Mean 49.3 51.7 53.2 55.1 51.4 Range 16.5-62.1 22.7-68.5 14.0-64.9 22.0-65.3 14.0-68.5 Index of coexistent disease 0 38 23 35 28 30 1 62 71 64 72 67 2or3 0 6 1 0 3 Unknown 11 3 6 1 21 Pretreatment prostate-specific antigen level Median, ng/dl 6.3 7.8 5.8 6.3 6.9 Mean, ng/dl 8.3 11.2 6.3 6.8 9.1 Range, ng/dl 0.8-70.3 0.9-77.4 1.3-18.3 1.1-14.3 0.8-77.4 10 ng/dl 81 68 93 78 77 10-20 ng/dl 14 23 7 22 17 20 ng/dl 5 19 0 0 6 Unknown 1 1 2 Gleason score 4-6 50 48 76 78 55 7 39 34 23 22 33 8-10 11 18 1 0 12 Unknown 2 1 3 Clinical stage T 1 75 75 78 100 T 2 25 25 22 0 Unknown 18 5 8 1 32 Risk category Low 34 35 55 56 39 Intermediate 32 32 22 44 31 High 34 33 23 0 30 Unknown 16 4 7 3 30 Abbreviations: RT, radiation therapy; SF-36, Medical Outcomes Study Short-Form Health Survey.

SYMPTOMS AFTER PROSTATE CANCER TREATMENT 3983 Table 3. Time Course of Unadjusted Urinary, Bowel, and Sexual Function Scales for 410 Patients With Early Prostate Cancer Who Chose Surgery, Radiation, or Initial Observation and Completed 24-Month Participation Time of Survey Baseline 3 Months 12 Months 24 Months Scale Score SD Score SD Score SD Score SD Urinary incontinence Radical prostatectomy 4.9 12.1 34.9 23.7 23.9 23.5 23.4 23.9 Non nerve sparing 4.8 10.8 43.3 23.5 30.2 23.5 30.6 24.5 1-sided nerve sparing 6.0 14.7 27.3 23.7 16.4 22.4 15.5 20.1 2-sided nerve sparing 3.5 10.2 27.9 18.6 19.6 21.9 21.2 24.0 External-beam radiotherapy 7.0 13.4 8.3 15.7 9.2 15.8 8.5 15.7 Brachytherapy 2.6 8.5 6.3 14.6 4.6 11.1 7.5 15.1 Initial observation 7.9 13.2 11.2 15.8 16.9 19.6 17.8 19.0 Urinary obstruction/irritation Radical prostatectomy 23.4 15.1 25.6 15.4 19.3 12.8 18.8 13.1 External-beam radiotherapy 13.1 15.2 17.6 16.4 13.0 15.2 12.1 15.3 Brachytherapy 7.1 8.9 15.8 17.1 9.0 10.2 11.1 11.8 Initial observation 23.1 11.0 23.1 10.8 19.8 8.4 24.4 14.1 Bowel problems Radical prostatectomy 4.9 8.1 5.4 7.0 4.4 5.9 4.8 6.0 External-beam radiotherapy 4.6 6.1 12.3 13.8 9.8 9.8 8.9 9.4 Brachytherapy 4.1 6.7 8.2 7.1 7.2 7.1 7.2 8.5 Initial observation 6.4 8.6 4.6 7.3 2.6 5.9 6.9 10.7 Sexual function Radical prostatectomy 23.9 32.4 77.9 25.2 73.7 25.4 68.5 27.4 Non nerve sparing 32.9 35.8 77.8 29.8 79.8 25.0 71.7 29.4 1-sided nerve sparing 17.4 27.7 81.1 18.4 71.0 25.6 67.6 27.0 2-sided nerve sparing 15.1 27.7 73.3 23.8 65.0 23.5 63.0 23.1 External-beam radiotherapy 51.2 37.4 65.5 34.6 65.8 32.0 69.2 32.3 Brachytherapy 30.1 32.3 40.4 34.5 42.4 35.6 45.0 33.1 Initial observation 37.3 35.4 31.2 33.5 43.0 32.8 50.7 34.3 Abbreviation: SD, standard deviation. modestly better at 24 months. At 3 months, only 17% of RP patients reported any erections in the past 4 weeks, and only 8% of men reported any erection firm enough for unassisted vaginal penetration, but scores rose to 41% and 20%, respectively, by 24 months after surgery. The ability to have orgasm, which is physiologically independent of erectile function, was disproportionately preserved, reported in 46% of all surgery patients at 3 months and 57% at 24 months. Despite older age, greater comorbidity, and worse pretreatment sexual function, EBRT patients sexual dysfunction was similar to that of surgery patients at 24 months after treatment. All treatment groups had increased sexual dysfunction at 24 months, including initial observation patients, although their dysfunction was first noted later, perhaps as a result of crossing over to active treatment. We summarize the time course of functional changes experienced by the major treatment groups in Figures 1 through 4. Predictors of Outcomes at 24 Months To address potential confounding, we evaluated pretreatment function, sociodemographic characteristics, and medical variables that differed by treatment group and could plausibly Table 4. Factors Associated With Sexual Dysfunction at 24 Months Factor Coefficient Standard Error P Model R 2 Sexual dysfunction at baseline 0.46 0.04.0001 36.9% Unilateral nerve sparing RP 15.70 5.02.002 Bilateral nerve sparing RP 13.83 6.34.03 Non nerve sparing RP 11.09 4.53.01 Brachytherapy 10.79 3.74.004 Age 0.52 0.22.02 Marital status 7.46 3.85.05 Intercept 4.83 14.53 Abbreviation: RP, radical prostatectomy. Fig 1. The time course of urinary incontinence, urinary obstruction/irritation, bowel problems and sexual function for patients who underwent bilateral nervesparing surgery.

3984 TALCOTT ET AL Fig 2. The time course of urinary incontinence, urinary obstruction/irritation, bowel problems and sexual function for patients who underwent unilateral nerve-sparing surgery. Fig 4. The time course of urinary incontinence, urinary obstruction or irritation, bowel problems and sexual function for patients who underwent external beam radiotherapy. influence 24-month outcomes. In univariate analysis (data not shown), both pretreatment scores and the primary treatment modality were strongly associated with 24-month scores for each symptom outcome. However, we encountered paradoxical results. For example, although increasing age was associated with increasing bowel problems and sexual dysfunction, younger age was associated with greater urinary incontinence, presumably because younger age is strongly associated with RP, which often causes incontinence. To adjust for potential confounders, we constructed multivariable models (Tables 4 through 6). For all outcomes, the strongest predictor was pretreatment function. Only treatment further influenced urinary incontinence. RP, especially nonnerve sparing, worsened incontinence. The likelihood of incontinence was much less with either radiation treatment, but bowel function (Table 6) was worse and declined with lower SF-36 Physical Component Scores. Age was significantly associated with sexual function in the multivariable model, as was being married, though only marginally. The association with marriage may reflect residual confounding, since surgery patients were more often married, or the possibility that a more readily available partner makes sexual dysfunction more evident. RP, whether nerve-sparing or not, was associated with a large increase in sexual dysfunction. Surprisingly, although 2-sided nerve preservation was associated with less increased dysfunction than 1-sided preservation, the increase was still less with non-nerve sparing surgery. BT patients had an improved sexual outcome, while outcomes of EBRT patients, comprising most of the remaining patients, were intermediate. DISCUSSION To make informed treatment choices and to anticipate both declining and improving treatment-related changes in function and to adapt to permanent symptoms, men need accurate, unbiased longitudinal information about the consequences of PC Fig 3. The time course of urinary incontinence, urinary obstruction or irritation, bowel problems and sexual function for patients who underwent non-nervesparing surgery. Fig 5. The time course of urinary incontinence, urinary obstruction or irritation, bowel problems and sexual function for patients who underwent brachytherapy.

SYMPTOMS AFTER PROSTATE CANCER TREATMENT Table 5. Factors Associated With Urinary Incontinence at 24 Months Factor Coefficient Standard Error treatment. Few studies have assessed treatment-related functional changes after early PC treatment over time, and all are limited. Our study s strengths include validated treatment-specific outcome measures responsive to change over time, detailed assessments of pretreatment function and other potentially confounding variables, and adequate longitudinal follow-up to identify long-term outcomes for most, if not all, domains. This study documents the characteristic consequences of each of three main treatment modalities. Because most men received treatment at a university-affiliated teaching hospital, adverse outcomes cannot be easily dismissed as a result of technically inferior care. The three treatment options we studied differed by the organ systems affected and the magnitude and time course of dysfunction. Radiation, especially BT, causes transient urinary obstruction/irritation, while bowel dysfunction after radiation, especially EBRT, and urinary incontinence after surgery improved, but persisted. All treatments caused sexual dysfunction, but when we controlled for other factors including pretreatment function, surgery had the greatest negative impact at 24 months, and nerve-sparing surgery had modest benefit at best. The fact that the time course of sexual dysfunction differs after surgery and radiation treatments is important for counseling patients and interpreting reported outcomes. Sexual function severely worsened immediately after surgery and then improved, while sexual function continued to decline after both radiation treatments. Thus, longer observation may reduce the sexual function disadvantage of surgery. Urinary incontinence after surgery did not improve after 12 months, and the modest improvement in bowel symptoms after external beam radiation therapy summates disparate trends among the individual symptoms comprising the function index. Bowel dysfunction increased after BT, a result not obvious in our earlier survey of patients treated a median of 5 years earlier. 22 The delayed decline in sexual dysfunction of the small group of patients choosing initial observation may be as a result of subsequent crossing over to active treatment, including androgen-deprivation therapy. P Model R 2 (%) Incontinence at baseline 0.57 0.07.0001 27.5 Non nerve sparing radical prostatectomy 17.51 3.09.0001 External-beam radiation therapy 8.68 2.12.0001 Brachytherapy 7.13 2.59.006 Intercept 13.35 1.68 Table 6. Factor Factors Associated With Bowel Function at 24 Months Coefficient Standard Error P Model R 2 (%) SF-36 physical component score 0.19 0.06.002 13.4 Bowel function at baseline 0.36 0.07.0001 Radiation therapy 2.42 0.90.008 Intercept 14.36 3.45 Abbreviation: SF-36, Medical Outcomes Study Short-Form Health Survey. 3985 To assist men in envisioning functional changes related to the alternative treatments, we provide figures that may serve as a functional road map. Clarifying the varied time courses of treatment-related complications may help patients better interpret their own status and assess prospects for future changes. A surgery patient who is incontinent 12 months later should consider improvement unlikely and perhaps entertain interventions to improve his condition, or begin adjustment to a condition likely to persist. A patient with bowel dysfunction 12 months after radiation therapy may expect modest improvement, with diverging trends for individual symptoms. Diarrhea, if any, will continue to subside, tenesmus and rectal urgency will change little, and episodes of rectal bleeding will become more prevalent. The surgery patient finding himself impotent at 3 or even 12 months after surgery can have some realistic hope of improvement; impotent radiation therapy patients probably should not. However, all men should be aware that erectile dysfunction, even if severe, does not preclude sexual satisfaction, including orgasm, and they should explore treatment options for sexual dysfunction with their urologist. The detailed pretreatment functional, sociodemographic, and medical information we collected allowed us to create a multivariable model to control, even if imperfectly, for the important differences between patient treatment populations that notoriously hinder understanding of PC treatment outcomes. Patientspecific factors such as age and pretreatment sexual function influence patient treatment choice, not only because they affect patient preferences, but also because they influence their physicians recommendations. Observational studies cannot determine with certainty whether different outcomes are as a result of treatment itself or pretreatment differences. However, while pretreatment symptoms strongly predicted all 24-month outcomes, underlining their importance in outcomes studies, the primary PC therapy had an important independent impact. Multivariable modeling could not completely eliminate confounding or render observed associations transparent. Older age protected patients in our study from urinary incontinence by making them less likely to undergo surgery, which causes it. Our multivariable model presumably did not find less sexual dysfunction after non-nerve-sparing surgery because of novel pathophysiology. Rather, patients who requested and whose surgeons attempted nerve preservation had much better pretreatment sexual function and thus greater potential functional decline, and the model incompletely adjusted for it. More broadly, our results display the difficulty of interpreting the net result of strong and sometimes opposing effects: factors influencing both the treatment received and the magnitude of its consequences. While consistent with the surprisingly limited sexual function benefit of nerve-sparing surgery we reported earlier after only 12 months, these results contradict our anomalous earlier finding of increased incontinence with the procedure. 10 These results may reflect the difficulty in accurately classifying surgical technique based on dictated operative reports that characterize the surgical attempt to preserve neurovascular bundles with variable accuracy. Further, physicians skill at nerve preservation may vary importantly. However, these results again suggest that benefits of

3986 TALCOTT ET AL nerve preservation may be overstated, even in the socially advantaged population seen at these hospitals. Our study has several limitations. Like nearly all published studies of early PC, our study is not randomized. Our rich data set allowed us to evaluate most potential confounders, although the limited racial and socioeconomic diversity in our population prevents us from assessing their impact. However, we found several factors that may influence the effects of treatment and which were strongly associated with the treatment received, limiting our ability to eliminate confounding, even in multiple regression modeling. The minimum 24-month follow-up allowed us to identify long-term dysfunction levels for urinary symptoms, but further follow-up may document additional changes in sexual and bowel function. Our follow-up period after treatment (as opposed to diagnosis, as in the otherwise exemplary Prostate Cancer Outcomes Study of the National Cancer Institute 4 ) is longer than any other large cohort study. The use of validated symptom scores produces precise, robust metrics for comparing patient groups and building models to address confounding. However, numerical results are difficult for patients and their physicians to interpret. To address this issue, we provide illustrative symptom frequencies to supplement the scale scores. Finally, our study does not document the broader impact of this dysfunction on the overall quality of these men s lives. Our validated bother scales for each of our urinary and bowel scales were highly correlated with our function scales, adding little to our measures of function. Additional measures assessing the impact of treatment-related dysfunction on patients lives are needed. These results confirm known associations between treatments for early PC and urinary, bowel, and sexual dysfunction, clarify other associations partially shrouded by confounding, and inform patients expectations for changes in function over time after treatment. We hope these results and others from the growing body of rigorous outcomes investigations will help patients better choose therapy and anticipate and adapt to dysfunction resulting from it, while helping treatment providers to assess their results and refine their techniques to improve them. ACKNOWLEDGMENT We thank Ms Sahirah Muhammad for assistance in preparation of the manuscript. AUTHORS DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST The authors indicated no potential conflicts of interest. 1. Litwin MS, Hays RD, Fink A, et al: Quality-of-life outcomes in men treated for localized prostate cancer. JAMA 273:129-135, 1995 2. Litwin MS, Flanders SC, Pasta DJ, et al: Sexual function and bother after radical prostatectomy or radiation for prostate cancer: Multivariate quality-of-life analysis from CaPSURE Cancer of the Prostate Strategic Urologic Research Endeavor. Urology 54:503-508, 1999 3. Litwin MS, Pasta DJ, Yu J, et al: Urinary function and bother after radical prostatectomy or radiation for prostate cancer: A longitudinal, multivariate quality of life analysis from the Cancer of the Prostate Strategic Urologic Research Endeavor. J Urol 164:1973-1977, 2000 4. Potosky AL, Legler J, Albertsen PC, et al: Health outcomes after prostatectomy or radiotherapy for prostate cancer: Results from the Prostate Cancer Outcomes Study. J Natl Cancer Inst 92:1582-1592, 2000 5. Stanford JL, Feng Z, Hamilton AS, et al: Urinary and sexual function after radical prostatectomy for clinically localized prostate cancer: The Prostate Cancer Outcomes Study. JAMA 283:354-360, 2000 6. Fowler FJ Jr, Barry MJ, Lu-Yao G, et al: Effect of radical prostatectomy for prostate cancer on patient quality of life: Results from a Medicare survey. Urology 45:1007-1013; 1995 (discussion 1013-1015) 7. Fowler FJ Jr, Barry MJ, Lu-Yao G, et al: Outcomes of external-beam radiation therapy for prostate cancer: A study of Medicare beneficiaries in three surveillance, epidemiology, and end results areas. J Clin Oncol 14:2258-2265, 1996 8. Litwin MS, McGuigan KA: Accuracy of recall in health-related quality-of-life assessment among men treated for prostate cancer. J Clin Oncol 17:2882-2888, 1999 9. Fowler FJ Jr, Barry MJ, Lu-Yao G, et al: Patient-reported complications and follow-up treatment after radical prostatectomy: The National Medicare Experience 1988-1990 (updated June 1993). Urology 42:622-629, 1993 10. Talcott JA, Rieker P, Propert KJ, et al: Patient-reported impotence and incontinence after nerve-sparing radical prostatectomy. J Natl Cancer Inst 89:1117-1123, 1997 REFERENCES 11. Talcott JA: Quality of care in prostate cancer: Important to start and too important to stop here. J Clin Oncol 21:1902-1903, 2003 12. Wasson JH, Cushman CC, Bruskewitz RC, et al: A structured literature review of treatment for localized prostate cancer: Prostate Disease Patient Outcome Research Team. Arch Fam Med 2:487-493, 1993 13. Robinson JW, Dufour MS, Fung TS: Erectile functioning of men treated for prostate carcinoma. Cancer 79:538-544, 1997 14. Beard CJ, Propert KJ, Rieker PP, et al: Complications after treatment with external-beam irradiation in early-stage prostate cancer patients: A prospective multiinstitutional outcomes study. J Clin Oncol 15:223-229, 1997 15. Talcott JA, Rieker P, Clark JA, et al: Patient-reported symptoms after primary therapy for early prostate cancer: Results of a prospective cohort study. J Clin Oncol 16:275-283, 1998 16. Greenfield S, Aronow HU, Elashoff RM, et al: Flaws in mortality data: The hazards of ignoring comorbid disease. JAMA 260:2253-2255, 1988 17. McHorney CA, Ware JE, Jr., Raczek AE: The MOS 36-Item Short- Form Health Survey (SF-36), II: Psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care 31:247-263, 1993 18. Clark JA, Talcott JA: Symptom indexes to assess outcomes of treatment for early prostate cancer. Med Care 39:1118-1130, 2001 19. Ware J, Kosinski M, Keller S. SF-36 Physical and Mental Health Summary Scales: A User s Manual. Boston, MA, the Health Institute, 1994 20. Clark JA, Rieker P, Propert KJ, et al: Changes in quality of life following treatment for early prostate cancer. Urology 53:161-168, 1999 21. D Amico AV, Whittington R, Malkowicz SB, et al: Biochemical outcome after radical prostatectomy, external beam radiation therapy, or interstitial radiation therapy for clinically localized prostate cancer. JAMA 280:969-974, 1998 22. Talcott JA, Clark JA, Stark PC, et al: Long-term treatment related complications of brachytherapy for early prostate cancer: A survey of patients previously treated. J Urol 166:494-499, 2001