A randomized trial of choice of treatment in prostate cancer: the effect of intervention on the treatment chosen

Authors


A. Auvinen, School of Public Health, University of Tampere, FIN-33014 Tampere, Finland.
e-mail: anssi.auvinen@uta.fi

Abstract

OBJECTIVE

To determine whether different approaches in the choice of treatment affect the treatment chosen by the patient for prostate cancer.

PATIENTS AND METHODS

We conducted a randomized trial with 210 men who had a histologically confirmed diagnosis of prostate cancer in 1993–94 at four major hospitals in Finland. After obtaining informed consent the men were randomized either to an intervention arm, in which there was greater patient participation in the choice of treatment following a structured procedure, or a control arm in which the standard approach, i.e. a standardized treatment protocol, was used. The main outcome measure of the analysis was the primary treatment chosen for prostate cancer.

RESULTS

In the enhanced participation arm patients not eligible for radical prostatectomy chose orchidectomy less frequently and favoured nonsurgical endocrine treatment than in the treatment protocol arm. Radical prostatectomy was the most commonly chosen treatment option in both arms among men with operable cancer. The way treatment options were presented affected the treatment chosen for prostate cancer.

CONCLUSION

Patients with prostate cancer are willing and able to take an active role in making decisions. The preferences of patients with prostate cancer in the choice of treatment may differ from the priorities of the physicians.

INTRODUCTION

Traditionally, the role of the physician in clinical decision-making has been that of an agent for the patient, using his or her knowledge for the best care of the patient. However, the autonomy of the patient requires that the patient is involved in decision-making [1]. For that purpose, the doctor should provide the patient with all essential information about the treatment options, possibly with a recommendation. This is especially important in cancer, as the disease is potentially lethal and many forms of treatment may have serious adverse effects.

However, it is unclear if similar treatment options are chosen regardless of whether a treatment protocol or active patient participation is emphasized in decision-making. Thus we conducted a randomized trial to assess how these approaches influence the treatment chosen for prostate cancer. This disease provides an ideal opportunity for the study because it is common, but there are several treatment options, none of which has so far been shown to be better than the others [2,3].

PATIENTS AND METHODS

The study protocol was approved by the institutional review board at each participating hospital. The consent form included information on the procedures and outcome measures, but the intervention was not described and the patients were unaware to which arm they belonged. All new patients with prostate cancer seen in the three central hospitals of the Tampere University Hospital District (Tampere University Hospital and Seinäjoki and Hämeenlinna Central Hospitals) and in the Kuopio University Hospital, and diagnosed with a microscopic confirmation between September 1993 and November 1994, were eligible for the trial. Only patients with an incidental diagnosis of latent prostate cancer (stage T1a), or unable to provide an informed consent because of dementia, or ineligible for treatment because of strongly impaired general condition, were excluded. A contraindication for one of the treatment options, e.g. severe coronary heart disease for radical prostatectomy, was not an exclusion criterion.

The target sample size was set at 200 patients; this enables the detection (α = 0.05, 1 − β = 0.8 using Fisher's exact test to compare two binomial populations) of a difference between arms in the proportion of patients selecting the most popular treatment options (orchidectomy in extraprostatic disease) of 0.9 vs 0.7, given that two-thirds of the patients have extraprostatic disease.

After explaining the purpose and procedures of the study, written consent was obtained. The participants were randomized centrally using a computer algorithm based on random numbers. The patients were randomly allocated into ‘enhanced participation’ and ‘treatment protocol’ arms. For logistic reasons (not having to interrupt the consultation), they were randomized before obtaining informed consent. The lag between randomization and intervention (patient encounter) was very short, typically < 30 min and only occasionally up to 24 h. Four urologists at four hospitals were in charge of the trial patients in both arms of the trial. They all were board-certified urologists with at least 10 years of clinical experience. No blinding was used, as the treatment chosen was recorded objectively (and therefore blinding was unnecessary).

In the ‘enhanced participation’ arm, the patient defined his own role in the choice of treatment. Systematic and comprehensive information on treatment options was given in a structured fashion, both orally and in writing. The patient and the physician discussed various aspects of available treatment options, including survival rate, adverse effects and cost, and the patient's opinion about the aims of treatment and willingness to accept potential side-effects. If the patient felt that he needed time to make a decision, or a family member to be involved, the choice of treatment was postponed. It was also possible for the patient to delegate the choice of treatment entirely to the urologist. This did not contradict the rationale of the study and it was not considered contamination (switching between trial arms), as it was one of the options in choice of treatment. The procedure was described in more detail elsewhere [4].

In the enhanced participation arm, the treatment options for patients with a localized disease (stage cT1b–T2, N0, M0), ≤ 70 years of age and with no other contraindications for surgery were radical prostatectomy, radiotherapy and watchful waiting. For patients with localized disease but aged > 70 years or with other contraindications for surgery, the options included radiotherapy, orchidectomy, LHRH agonist, antiandrogen, oestrogen and watchful waiting. For patients with extracapsular disease (cT3–T4), lymph node involvement (N1–N3) or distant metastasis (M1), the options were orchidectomy, LHRH agonist, oestrogen and antiandrogen.

In the control arm, standard treatment protocols were used and the option regarded as the treatment of choice was offered as the primary option. Relevant aspects of treatment, including the procedures and possible adverse effects, were discussed and other treatment options briefly described. The choice of treatment was based on informed consent and the patient had the option of refusing the treatment of choice. For patients with localized disease, aged up to 70 years, and with no other contraindications for surgery, radical prostatectomy was recommended as the treatment of choice. Orchidectomy was the primary option for men with advanced cancer and for those for whom surgery was contraindicated.

In the univariate analysis, the statistical significance of the differences was assessed using the exact Pearson chi-square test. Multivariable logistic regression analysis was used to control for the effects of age, grade, hospital and pretreatment serum concentration of PSA on the chosen treatment in the two arms of the trial.

RESULTS

In all, 210 patients of 251 fulfilling the eligibility criteria were randomized, 104 into the enhanced participation arm and 106 to the treatment protocol arm (Fig. 1). Seven randomized patients were excluded after randomization; four for refusal (randomization was before consent) and three for rapid deterioration of health (progressive cancer or concomitant disease). These seven patients were included in the analyses (intention-to-treat principle).

Figure 1.

A flow chart of the recruitment and allocation in the randomized trial of choice of treatment in prostate cancer.

Forty-one eligible patients (16%) could not be enrolled as they were not seen by a study urologist. In terms of baseline characteristics, their age distribution was comparable with that of the study participants (mean age 72 vs 71 years), but they more often had extraprostatic disease (82% vs 66%). Of the 41, 35 (85%) received endocrine treatment, five (12%) had a radical prostatectomy and one (2%) was treated with radiotherapy.

Randomization resulted in a balanced distribution of most major prognostic factors between the trial arms (Table 1), but the proportion of men with a serum PSA level of > 1000 ng/mL was higher in the treatment protocol arm (3% vs 7%).

Table 1.  Characteristics of the patients by age, stage and trial arm in randomized total of choice of treatment in prostate cancer
CharacteristicEnhanced, n (%)Standard, n (%)
  • *

    Mostofi grade; grade I corresponds roughly to Gleason grade 2–4, II to Gleason 5–7, and III to Gleason 8–10;

  • † Treatment of choice recommended for the treatment protocol group.

Age group, years
≤59  7 (7)  3 (3)
60–6944 (42)37 (35)
70–7939 (38)45 (42)
≥8014 (13)21 (20)
Karnofsky score
10046 (44)54 (51)
9029 (28)28 (26)
60–8026 (25)20 (19)
40–50  1 (1)  1 (1)
Missing  2 (2)  3 (3)
WHO grade*
I27 (26)33 (31)
II58 (56)59 (56)
III19 (18)14 (13)
PSA range, ng/mL
0–3.9  4 (4)  6 (6)
4–9.917 (16)16 (15)
10–9972 (69)69 (65)
100–999  8 (8)  8 (6)
≥1000  3 (3)  7 (7)
Stage
Intracapsular39 (38)32 (30)
Locally advanced45 (43)52 (49)
Distant20 (19)22 (21)
Choices
Local disease, aged ≤ 70 and no contraindications for surgery
Watchful waiting  2 (7)  2 (11)
Radical prostatectomy17 (63)15 (83)
Radiotherapy  8 (30)  1 (6)
Total27 (100)18 (100)
Local disease, aged > 70, or ≤ 70 with a contraindication
Watchful waiting  8 (67)  5 (36)
Radiotherapy  2 (17)
Orchidectomy  1 (8)  7 (50)
LHRH agonist  1 (8)  2 (14)
Total12 (100)14 (100)
Not local disease
Watchful waiting  1 (2)  3 (4)
Radiotherapy  2 (3)
Orchidectomy42 (65)69 (93)
LHRH agonist19 (29)  2 (3)
Total androgen blockade  1 (2)
Total65 (100)74 (100)

Overall, 58% (60/104) of the men in the enhanced participation arm selected the option defined as the standard treatment (treatment of choice), vs 86% (91/106) in the treatment protocol arm (difference in proportions 28%, 95% CI 15–42%, P < 0.001). Among patients eligible for surgery (local stage, ≤ 70 years old) most were treated with radical prostatectomy in both arms (83% vs 63%, difference in proportions 20%, 95% CI − 11 to 50%, P = 0.19; Table 1).

Among patients for whom surgery was contraindicated because of comorbidity or age, but with local disease, those in the enhanced participation arm were less likely to choose orchidectomy than in the treatment protocol arm (50% vs 8%, difference in proportion 42%, 95% CI 3–77%, P = 0.04, Table 1).

Among patients with advanced disease, 29% of the subjects in the enhanced participation group chose treatment with LHRH agonist, but only two did so in the treatment protocol group (for the LHRH agonist and total androgen blockade group combined, 31% vs 1%, difference in proportions 29%, 95% CI 18–42%, P < 0.001). In this group orchidectomy was the treatment most commonly (93%) chosen in the treatment protocol arm, whereas there was more variation by treatment in the enhanced participation arm (93% vs 65% for orchidectomy, difference in proportions 29%, 95% CI 14–45%, P < 0.001, Table 1).

The differences in types of treatment chosen between the arms among patients with inoperable disease (not organ-confined cancer or age > 70 years) were unaffected by adjustment for grade, pretreatment PSA or Karnofsky performance status in the multivariate logistic regression analysis (crude odds ratio for endocrine treatment, 2.2, 95% CI 0.9–5.5, adjusted odds ratio 2.2, 95% CI 0.8–5.9 in treatment protocol vs enhanced participation arm). Similarly, adjustment for these factors had no substantial effect on differences between the trial arms among subjects with operable disease (crude odds ratio for radical prostatectomy 2.6, 95% CI 0.6–12, adjusted odds ratio 3.2, 95% CI 0.5–20 in treatment protocol vs enhanced participation arm).

DISCUSSION

This is to our knowledge the first randomized trial comparing two approaches to patient participation in the choice of treatment in prostate cancer. Enhanced participation resulted in more frequent non-surgical endocrine treatment among men with inoperable prostate cancer than in men in the treatment protocol arm. Radical prostatectomy was the most commonly chosen form of treatment for operable prostate cancer in both arms of the trial.

These results show, first that there is no single treatment approach that is optimal for all men, and second that the treatment most widely favoured by the patients is not the same as that favoured by the clinicians in charge of treatment. We consider these results to be relevant not only for these treatment options but also more generally for men with prostate cancer, regardless of treatment policy, because our intervention dealt with the approach to decision-making, rather than any specific form of treatment.

Radical prostatectomy was also the most commonly chosen treatment option in an earlier descriptive study of treatment preferences among men with localized prostate cancer [5]. In extraprostatic prostate cancer a previous descriptive study also reported a preference for chemical rather than surgical castration among most men with advanced prostate cancer, and there has recently been a shift towards a chemical rather than surgical approach in endocrine treatment [6,7]. No differences in survival have been detected between surgery and radiotherapy in localized prostate cancer, nor between various forms of endocrine treatment in advanced disease. Furthermore, recent studies also suggest that the frequency of adverse effects and quality of life outcomes are comparable for different treatment options [8]. Preference for non-surgical treatment among men with inoperable prostate cancer may reflect more convenient administration or reversibility of treatment, or psychological effects of maintaining the testicles (e.g. perception of masculinity).

In the enhanced participation arm, a third of the men preferred the physician to make the treatment decision. In the treatment protocol arm 90% of the patients accepted the recommended treatment. However, a limitation in the assessment of this secondary endpoint was that it was done by the physician responsible for the intervention, because a blinded evaluation was not feasible. Previous descriptive studies have also found that most patients with cancer expect a treatment recommendation from the physician and are willing to accept it, provided that their information needs are met [9–12].

We were unable to enrol a sixth of the eligible patients; even though a slightly larger proportion of them were diagnosed with advanced disease than in the randomized patients, we do not think this diminishes the external validity of the results. This is further confirmed in that the stage distribution in the source population for the trial (including men ineligible because of poor general condition) was comparable with that among all prostate cancer patients in Finland (70% vs 74% not localized disease) [13]. Failure to enrol all eligible patients was a consequence of the trial design, i.e. only one urologist per centre was involved in the trial and was responsible for treating patients in both arms. This approach was chosen to eliminate the effect of the physician in charge of treatment. Indeed, we found it preferable a priori to have a proportion of withdrawals with all the randomized patients being treated exclusively by physicians familiar with the study and committed to it, rather than pursuing complete enrolment with imperfect implementation of the protocol in some cases. However, this resulted in inability to enrol patients not seen by participating urologists, even though they were usually in charge of treating most of the patients with prostate cancer at the hospital. Randomization led to a slight imbalance between the arms in pretreatment serum PSA levels, mainly attributable to a few very high values. However, multivariate analysis with adjustment for PSA level and other factors confirmed the results of the univariate analysis.

The choice of treatment ultimately depends on preferences or relative values (utilities) given to length of life, quality of life and cost of treatment. More individualized treatment has been called for, as the preferences may differ both between the physician and the patient, and among patients [14–16]. There is empirical evidence for a difference in preferences related to aims and side-effects of treatment among patients [6,12,17–19]. Among patients with prostate cancer the value that they give to certain aspects of their health, including those affected by the disease and different treatment options, is affected by age, health status and psychological features [17,20]. Some patients may be willing to trade life-years, e.g. to ensure maintained sexual functioning [17]. In addition, priorities may differ between patients and physicians. The issues regarded as relevant for decision-making in the choice of treatment differ among patients and physicians [15,18,21]. Treatment can be tailored according to individual preferences only by means of active patient participation. The present results, and those in previous studies, show that instead of mere compliance, most patients with cancer are willing to take an active role in decision-making [5,10–12]. Recommendations for a shared decision-making approach have been incorporated in most recent treatment guidelines [22]

There are several treatment options in both localized and advanced prostate cancer, none of which is clearly better, and there are potentially important side-effects, e.g. impotence and urinary incontinence, associated with treatment. Therefore, the decision-making situation may be regarded as a matter of chance [23]. However, we consider that these results are also applicable to patients with other diseases, especially in situations where no treatment option is confirmed to be better than the others. The differences in treatments between the arms were most pronounced among patients with advanced disease. As there is currently no curative treatment available for these patients, the choice of treatment probably reflects even more the compromise between quantity and quality of life.

We randomized more than 200 patients, almost all of whom agreed to give informed consent. There were statistically significant differences in the distribution of patients by treatment between the intervention arms. This implies that the patients were able and willing to participate in the clinical decision-making. Furthermore, the results suggest that patients may not have the same preferences as those on which the current treatment protocols are based.

ACKNOWLEDGEMENTS

The study was supported financially by the Finnish Cancer Institute, Academy of Finland, Cancer Society of Finland, Pirkanmaa Cancer Society and Pohjois-Savo Cancer Fund. We thank the study nurses Eeva-Liisa Härkisaari, Liisa Kuivalainen, Leila Lakso, Seija Rintala and Maritta Salminen for data management; Marita Hallila for secretarial assistance; Bengt Söderman for programming; and Marja-Liisa Weckman for data entry.

EDITORIAL COMMENT

This is a very interesting paper, which discusses topical issues about direct patient involvement in decisions of treatment choice. It builds upon excellent work already published by the same authors in the BJU International. However, several points in this study might provoke criticism. The technique of randomization which was conducted before informed consent is controversial, although it is probably acceptable in the context of this study. Similarly, significant variability might be expected between individual clinicians and their approach to patients, despite protocols being described to minimise such variations. It might also be argued that the conclusions of the study are obvious and entirely as might be expected. Patients will often choose different treatments to those ‘preferred’ by their doctors, especially when surgical treatment vs endocrine therapy for advanced prostate cancer is concerned in units where orchidectomy is offered as the treatment of choice. However, overall the paper is well written and referenced, and serves as a timely reminder that all clinicians should strive to involve their patients more in decisions about their medical management.

J. NOBLE
Department of Urology, Churchill Hospital, Oxford, UK

Ancillary