Influence of Radioimmunoscintigraphy on Postprostatectomy Radiotherapy Treatment Decision Making

MATERIALS AND METHODS

The patient population under study for this investigation was the group of post-RRP prostate cancer patients appearing for consultation in our hospital consortium between 1998 and 2002. These years were selected because, although post-RRP RT has been done within our consortium for many years, RIS first became available and was first clinically implemented in our consortium in the post-RRP setting in 1998. The charts of 54 consecutive patients having prostate cancer who (a) underwent RRP, (b) were referred to our hospital consortium for consideration of external-beam RT for biochemical failure post-RRP or for high-risk of failure post-RRP, and (c) had an RIS scan ordered for aiding RT decision making were reviewed. The database used for this investigation was approved by the Institution Review Boards of all of the hospitals whose patient data were used for this investigation. Because this investigation was retrospective, a formal waiver of informed consent was requested and approved before conducting the study.

The general demographics and pre-RT treatment and follow-up history were reviewed. The characteristics of these patients are shown in Table 1, which displays the patient age, RRP pathologic findings (stage, grade, margin status, seminal vesicle invasion status, extracapsular extension, and lymph node status), postprostatectomy course leading to RT consultation (post-RRP PSA nadir, post-RRP PSA follow-up course, interval from RRP to RT consultation, and administration of hormones), and post-RT follow-up information. In a very few patients, due to the long interval between surgery and RT consultation, the original prostatectomy pathology report was unavailable for tabulation of the pathologic findings; these are recorded as “uncharted” in Table 1. As displayed, there are several patients with very low pre-RT PSAs that may not have expected to impact the study greatly. Although one might at first consider excluding these patients, thus allowing analysis of patients expected to have a higher yield of RIS-based decision changes, it was important to include these patients to preserve the consecutive nature of the study.

The RT recommendations before knowledge of the RIS findings were reviewed with regard to (a) the decision to recommend RT and with regard to (b) the general target volume to be treated (i.e., whether to deliver RT only to the PF or whether to deliver RT to the PF and the pelvis [PF+P] [defined as treatment to a pelvic volume larger than the PF, including unilateral or bilateral pelvic lymph nodes]). This review was feasible because the RIS scan was ordered by the radiotherapist or referring urologist at or near the time of the RT consultation. In the majority of cases, the RIS scan was ordered by the radiation oncologist at the time of initial consultation, so the initial plan was recorded before knowledge of the RIS information. In other cases in which the RIS scan was ordered by the referring urologist before the RT consultation, the official RIS reading was often unavailable at the time of consultation. In a few instances, the result of the RIS scan was available before the RT consultation, and even in these cases it was possible to infer from the consultation dictation or record what the RT plan would have been without the RIS information. Thus, it was possible in all cases to chart the RT plan before the knowledge of the RIS findings.

Of note, the most commonly used radiologic scans—the bone scan and CT scan of the abdomen or pelvis—were ordered in 20 and 15 patients, respectively, immediately before or at the time of RT consultation, and in all cases these tests were negative. The intent in all cases at the time of radiation referral was to offer RT with curative intent, as the metastatic work-up (when performed) and the pace of PSA rise was, before obtaining the RIS scans, believed to be consistent with disease localized to the pelvis. Although no uniform guidelines are yet established or adhered to (in the RT community in general or in our particular institution consortium), in general, PF radiation was thought to be indicated for patients with low-grade pathology, low PSA nadir, and long interval between nadir and failure; these criteria applied to the vast majority of patients in the current study. PF+P was recommended primarily because of aggressive pathology (Gleason score, >8 with seminal vesicle invasion) and corresponding increased lymph node risk. The absolute PSA (pre-RRP or the highest value pre-RT) was generally not a significant factor in determining the RT to the pelvis in our hospital consortium; this may be subtly different from practices at other institutions. Review of the initial (pre-RIS) RT recommendations revealed that, of the 54 total patients, the RT decision was to treat the PF only in 52 patients and to treat the PF+P in 2 patients.

The RIS reports were then reviewed on each of these patients and summarized. The RIS scans were read by one board-certified nuclear medicine physician who was involved in the initial development of the procedure as well as in several RIS clinical trials. Because only one physician read the RIS scans, interobserver variability in interpretation of the scans is not a confounding variable in this particular investigation. Planar and volume SPECT datasets were obtained using a dual-head SPECT Prism 200 Picker camera (Philips). This dual-energy procedure for acquisition of data and interpretation is described in considerable detail elsewhere (16) and thus is not repeated here. The scans were read with knowledge of the patient’s clinical history but not with the aid of CT or MRI information; in addition, a CT/SPECT system was not used. For each patient, the RIS findings, with regard to uptake in the PF, uptake in the P (i.e., uptake within the pelvis in a region outside of the PF), or EP uptake were reviewed.

Finally, the RT treatment decisions based on the RIS findings were reviewed for each patient. Any change in the recommendation to offer RT and any change in the general target volume to be treated were noted. Thus, there were 3 categories of final RT decisions: (a) no RT, (b) RT to PF only, and (c) RT to PF+P. The decision to abort RT was generally based on the RIS scan showing uptake in EP regions or the pattern of pelvic uptake (intensity and level of pelvic involvement [i.e., solitary vs. matted or multiple sites within the pelvis], as the risk for occult EP disease may rise). The decision to offer RT to the PF+P (changed from the decision to offer RT to the PF alone), was usually undertaken if the RIS scan showed uptake in the pelvis but not strong enough uptake to warrant consideration of occult EP disease. In each case, the provider reviewed the RIS findings in the clinical context of the patient’s pathology, PSA nadir, disease-free interval, pace of PSA rise, CT or bone scan findings [if available], pre-RIS decision, and the known sensitivity or specificity rates of RIS to determine whether RIS would override the established pre-RIS decision. Because the study was retrospective, spanning several hospitals, each with several providing physicians, and because no specific published guidelines existed (or exist to date) in the literature on how to incorporate RIS into RT decision making, stylistic variations were expected to exist among the 8 different practitioners represented in this study and among the individual patients in their level of desire for aggressive management of the prostate cancer weighed against the potential side effects of RT. Within this framework, however, each individual provider was usually consistent in the approach.

Although the treatment technique and dose delivered are not critical to the analysis of the decision making in this investigation, they do bear brief description: In category (b) of the final RT decision, the PF dose typically used at our institution was 66.0 Gy (in 2.0-Gy, once-daily fractions), and in category (c), the regional pelvic lymph nodes were treated to an initial dose of 50.4 Gy (in 1.8-Gy, once-daily fractions) or 50.0 Gy (in 2.0-Gy, once-daily fractions) followed by a PF boost to 16 Gy (in 2.0-Gy, once-daily fractions) to achieve a final dose of 66.0–66.4 Gy. The final dose used was based on the American Society for Therapeutic and Radiation Oncology consensus conference (22), which recommended a minimum dose of 64.0 Gy. When the pelvic lymph nodes were treated, a 4-field technique was typically used. The prostate bed was treated with 6-field conformal therapy between 1998 and 2000 and was treated with intensity-modulated RT from 2000 onward.

A Kaplan–Meier curve (23) was generated for biochemical failure-free survival, based on available post-RT follow-up PSA information. The definition of failure used was the presence of 2 consecutive PSA rises above the level of 0.2 ng/mL after reaching a nadir; patients were also declared to fail after 2 rises if no nadir was reached. This definition of failure combines features of definitions relying on successive rises and definitions relying on an absolute PSA threshold (7–10). In addition, univariate and multivariate analyses of patient, disease, and treatment factors were performed using the log rank test and Cox proportional hazards regression, respectively (23).

RESULTS

Table 2 shows the results of these analyses. Displayed for each patient are the individual RT provider, the RT decision before knowledge of the RIS findings, the RIS findings, and the RT decision after the knowledge of the RIS findings. Of the 54 patients originally referred for prostate RT, the initial decision was to recommend RT in all 54. In 2 cases (patients 24 and 36), the initial plan was to treat the PF+P; for the remaining 52 patients, the initial decision was to deliver RT to the PF only.

As shown in Table 2, the RIS findings were as follows: PF only in 43 patients, PF+P in 8 patients (patients 17, 20, 21, 25, 30, 37, 45, and 53), PF+EP in 2 patients (patients 11 and 22), and PF+P+EP uptake in 1 patient (patient 54).

As shown by the listing of post-RIS decisions in Table 2, after knowledge of the RIS results, the decision to withdraw the RT recommendation was made in 4 patients (patients 22, 30, 45, and 54). This decision was made due to pelvic uptake in 2 patients (patients 30 and 45), EP uptake in 1 patient (patient 22), and both pelvic and EP uptake in 1 patient (patient 45). Of the 54 patients for whom RT was initially recommended, RIS changed the general treatment volume in 6 patients (patients 17, 20, 21, 25, 37, and 53). This decision was due to pelvic uptake outside of the prostate bed in all 6 patients, causing the radiotherapist to design a more generous volume that encompassed the regional lymphadenopathy seen on RIS or, in most cases, a more generous volume to include the whole pelvis.

The influence of the RIS findings and a systematic comparison of the pre-RIS and post-RIS RT decisions are summarized in Table 3:

The top part of Table 3 displays the results of the analysis of the influence of RIS on the decision to offer RT. After knowledge of the RIS results, the decision to withdraw the RT recommendation was made in 4 of the 54 patients (7.4%). Although the χ² statistic is used when comparing proportions in 2 populations and was computed, the simple χ² statistical test is not entirely appropriate for evaluation of the significance level, as the 2 populations being compared are not independent. Thus, the significance level was evaluated using the McNemar test (24), which is a standard statistical test used for comparing the proportions of 2 dependent populations. As displayed in this part of Table 3, the χ² statistic is 4.0 (with 1 degree of freedom), and this reached statistical significance (P = 0.045).

The middle part of Table 3 displays the results of the analysis of the influence of RIS on the specific RT treatment field recommendations. In 6 of the 54 patients (11.1%), the knowledge of the RIS findings resulted in changing the treatment volume from PF to PF+P. These findings were analyzed in a manner similar to that of the top part of Table 3, again using the McNemar test. As displayed, the χ² statistic was 6.0 (with 1 degree of freedom), and this also reached statistical significance (P = 0.015).

Finally, the bottom part of Table 3 displays the results of the overall influence of RIS on the RT recommendations. As discussed earlier, knowledge of the RIS findings altered the RT decision a total of 10 times (a treatment volume change in 6 cases and the decision to withdraw RT altogether in 4 cases), in 54 patients (10/54 = 18.5%). This part of Table 3 displays the overall decisions pre- and post-RIS. Again, as for the top and middle parts of Table 3, although the χ² statistic was computed, it was not appropriate to test the significance level using the simple χ² test. Furthermore, because in this case there are 3 possible outcomes (RT to PF, RT to PF+P, and no RT), the McNemar test was also not entirely appropriate. Assuming that all 3 decisions are equally weighted, the statistical test of choice was the Stuart–Maxwell test (25,26), which is a standard test used to compare nondichotomous decisions in 2 dependent populations. (The Stuart–Maxell test actually reduces to the McNemar test used for the top and middle parts of Table 3 when the decision is dichotomous.) As displayed, the χ² statistic was 10.0 (with 2 degrees of freedom), and this also reached statistical significance (P = 0.0067).

Figure 1 displays the Kaplan–Meier curve for the entire cohort; as shown, the 3-y biochemical failure-free survival was approximately 78%. Table 4 shows the results of the univariate and multivariate analyses, which were performed using the listed patient, disease, and treatment factors as covariates. As displayed in Table 4, no factors demonstrated statistical significance. Only pathologic T-stage, highest pre-RT PSA, and hormone therapy demonstrated trends on the univariate analysis; on further multivariate analysis, even these parameters did not display statistical trends.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Kaplan–Meier biochemical failure-free survival curve.

DISCUSSION

RIS directed against PSMA was found in this investigation to influence postprostatectomy RT decision making for the identification of patients with disease not likely to benefit from RT and for guiding general RT treatment volume definition. This study demonstrated that in the select group of patients who have a rising PSA post-RRP or are at high risk of failure post-RRP, and in many cases having negative bone scan or abdomen or pelvis CT scan, the RIS scan provides additional information that is useful in guiding the RT decision making. This was demonstrated in 2 ways: (a) altering the decision to offer RT and (b) altering the decision to offer RT to a different target volume than initially intended. These results might be expected, because RIS has a different mechanism of identification of the disease and the information it provides would be expected to be somewhat complementary to the current tools currently available to guide RT decision making. However, to our knowledge, this investigation is the first report to demonstrate a statistically significant change in RT decision making based on RIS findings.

The number of changes in the RT decision making is a strong function of the yield of the RIS findings in the patient population under study. In this context, the expected rates of uptake in different anatomic regions at the time of RIS are relevant. The key multiinstitutional study (15) sheds light on this matter, and the rates of RIS uptake in the setting of post-RRP in that study were as follows: PF, -65%; PF+P, -25%; PF+P+EP, -20%; EP, -10%. Our investigation showed a higher percentage of PF-only uptake (42/54 = 78%) and a corresponding lower percentage of P and EP uptake, due in part to the fact that bone scans and CT scans may have excluded some individuals who also would have had RIS uptake in the P and EP regions. Thus, it is probable that if the RIS scans were done on a de novo population to guide RT (i.e., a population in which no prior bone scan or CT scans were obtained), the percentage uptake on the RIS scans would be higher than that of the current investigation. Consequently, the RT decisions would likely have been altered more frequently, and the conclusions of the current investigation would likely be further strengthened.

The effect of the referral pattern on interpretation of the results warrants discussion. In particular, there is expected to be a strong referral bias, as the current investigation only evaluates those patients who were referred by urologists for consideration of curative RT. Patients who had positive bone scans, positive CT scans, or even positive RIS scans before the RT referral were excluded from this study and therefore cannot be addressed by the current analysis. However, although the results of our study may not be relevant to the general group of post-RRP patients, the results are relevant to the practicing radiotherapist, as the patient population in our study is similar to that seen in most RT clinics, and our study addresses the use of RIS in assisting decision making in this specific setting.

Although the RIS information was useful in the vast majority of patients in this study, there were instances in which the RIS findings were not fully incorporated into the final decision, resulting in several seeming inconsistencies that bear discussion. Specifically, in 2 patients (Table 2, patients 24 and 36), the original plan was to deliver RT to the prostate and regional lymph nodes. This plan was carried through, despite the RIS scan showing uptake in the PF only. Thus, the RIS scans in these cases were believed to represent false-negative findings in the pelvis in both cases, and the RT clinician opted to make the decision, based on the rise in PSA and post-RRP findings, to continue with the original plan. In these cases, RIS was, however, useful as an additional tool (similar to the bone scan and the CT scan of the abdomen or pelvis) to rule out EP disease. Conversely, although EP uptake was demonstrated in patient 11, this was not thought to be consistent with the patient’s clinical picture (the EP uptake was in the supraclavicular fossa, a site amenable to physical examination), and the RIS reading was viewed as a false-positive finding; in other instances of EP uptake (patients 22 and 54), the location of the EP uptake was intraabdominal, and the RIS findings were read as true-positives. Finally, there are seeming inconsistencies in the use of the RIS scan showing PF+P uptake; in some cases (patients 17, 20, 21, 25, 37, and 53), RIS uptake in the PF+P led to a decision to treat the PF+P, whereas in other cases (patients 30 and 45), uptake in the PF+P caused a decision to abort RT altogether. These seeming inconsistencies are due in part to the pattern of uptake in the PF+P (and relative concern for occult EP disease) and in part to provider bias. In summary, although RIS did influence the RT decision making significantly in this investigation, there were instances in which clinical judgment did override the RIS findings. Many of the seeming inconsistencies can be explained by false-positive and false-negative findings, the pattern of RIS uptake, and variations in clinical judgment. Although there are expected variations between different providers, each provider was usually consistent in the decision making. Because no institutional policy existed or has been reported in the published literature on the incorporation of RIS for RT decision making, our results are likely a representative cross section of those expected in the general RT community. Of important note, the RT provider did not reach significance on univariate analysis (Table 4), suggesting that, within the limitations of the power of the study (as there were numerous radiation oncologists participating in the study), provider bias does not confound interpretation of the biochemical control outcome reported in this investigation.

Even with knowledge of the variations among providers, it is nonetheless important to quantify the findings to determine the general influence of RIS in this study population, which is what Table 3 sought to answer. Table 3 and the corresponding statistical tests account for the cases in which the RIS scan was ignored as well as when it did influence the decision: The off-diagonal elements represent a positive influence, whereas ignoring the test causes greater weighting of the diagonal elements and weakens the role of the RIS scan. The statistical test used in the top part of Table 3 assumes equal weighting of the decision to withdraw RT and the decision to offer RT, the statistical test used in the middle part of Table 3 assumes equal weighting of the decision to offer RT to the PF only and to offer RT to the PF+P, and the statistical test used in the bottom of Table 3 assumes equal weighting of the decision to withdraw RT and the decision to offer RT to a different volume than originally intended. Although modeling these weightings poses its own set of challenges, if the decision not to offer RT (a fundamentally major change in treatment course) or to offer RT to the PF+P is weighted higher than the decision to offer RT to PF, the off-diagonal elements in Table 3 would be weighted greater and the current study would reach even higher statistical significance than that currently reported.

Independent of these observations about statistical tests used to determine the influence of RIS on decision making, a key contribution of this investigation for the practicing clinician is the biochemical control analysis. The biochemical failure-free survival rate in the current investigation is, with available follow-up, similar to or higher than most reported postprostatectomy RT series (7–10). As shown in Table 4, no patient, disease, or treatment factor reached statistical significance. It is noteworthy that, in the current investigation, the interval between prostatectomy and RT (a surrogate for immediate RT vs. salvage RT) and hormone therapy did not reach significance, suggesting that these factors did not impact RIS uptake and the resulting treatment decisions strongly enough to influence the survival outcome. It is also noteworthy that the RIS findings themselves did not affect the survival. Correlating RIS findings with survival outcome has been undertaken by other investigators, with both positive (18) and negative (19) studies having been reported. It is important to understand that our investigation was not designed to answer this particular question, and the lack of significance of the RIS findings on univariate analysis should not be interpreted as a negative finding for the influence of RIS on survival, as in our study the survival detriment in having RIS uptake outside the PF was likely negated by the survival advantage in designing, in most cases, larger RT portals to encompass these areas of uptake.

The results of this study can form the basis for undertaking similar studies examining the role of RIS in the setting of pre-RRP decision making, in the setting of guiding decisions related to seed placement when performing a brachytherapy implant, and for decision making regarding selection of therapy in the setting of recurrent disease after radical RT.

In addition, the results of our study can be expanded to further analyze the role of RIS even within the current study population. First, the true survival advantage of using RIS to assist in decision making would require the study of a matched-pair cohort not undergoing RIS, as the patients in the current study can serve as their own controls for the decision-making process but not for the treatment outcome analysis. Second, the potential to use the exact regions of uptake on the RIS scan to guide the definition of the PF target bed, as an extrapolation of the work done in registering the RIS scan to the planning CT scan (27), has been studied (28).

The biases and limitations inherent in a retrospective review are understood by the investigators. Within these limitations, however, the current investigation sheds light on the role of RIS in the setting of post-RRP RT decision making as a tool to aid in selecting the patients most likely to benefit from RT and to aid in designing the general radiation field arrangement. It is hoped that the results of this investigation can guide prospective investigations in this area.

CONCLUSION

RIS was found in this investigation to influence post-RRP RT decision making for the identification of patients not likely to benefit from RT and for guiding general target volume definition. Our study can serve as a framework with which to design prospective trials in this area of investigation.