Caveat Emptor - European Urology

32 downloads 13403 Views 192KB Size Report
adjuvant or salvage therapy, since no postoperative PSA data exist. Multimodal therapy postoperatively may not be associated with worse cancer control and ...
EUROPEAN UROLOGY 66 (2014) 673–676

available at www.sciencedirect.com journal homepage: www.europeanurology.com

Platinum Priority – Editorial and Reply from Authors Referring to the article published on pp. 666–672 of this issue

Caveat Emptor R. Jeffrey Karnes a,*, Steven Joniau b, Michael L. Blute c, Hendrik Van Poppel b a

Mayo Clinic, Rochester, MN, USA; b University Hospital, Leuven, Belgium; c Massachusetts General Hospital, Boston, MA, USA

Although the purpose of the study by Hu et al. [1] in this issue of European Urology was not to sell the da Vinci robot, the study will probably be used by some to market it. We believe caution should be exercised when interpreting this study and its implications about whether this tool can lead to better prostate cancer (PCa) control outcomes. We should not readily believe the data being ‘‘sold’’ in this paper. Those who do not use or own the robot should not rush out to buy it. The tool does not make the carpenter. This study comes from an esteemed group of investigators in population-based urologic research and is not being published as a selling tool but rather as data reported. Conclusions are only as strong as the data, and this is where we feel this paper falls short. The US Surveillance Epidemiology and End Results (SEER)–Medicare data used in this study included men undergoing radical prostatectomy (RP) between 2004 and 2009. In the latter period of these data, robot-assisted RP (RARP) became the predominant form of RP in the United States. After close to 9000 exclusions (more than one-third), primarily for missing data (stage, grade, prostate-specific antigen [PSA]), an initial study group was developed of 5524 RARPs and 7878 open RPs (ORPs). After 1:1 propensity-based matching, an equal number of ORPs were created (n = 5524). With RARPs, there were lower PSA values, lower clinical stages, and lower grades of disease. In our opinion, this is suggestive of selection bias, and hence propensity-based matching was used. It is not known what the data were regarding positive surgical margins (PSMs) or secondary treatments (androgen-deprivation therapy [ADT] and/or radiation therapy [RT]) for the entire ORP cohort. Exclusions also included >1200 men with pT3b, arguably one of the most important unfavorable pathologic variables. This represents many data excluded.

PSMs were less frequently recorded in RARP, at 13.6% versus 18.3% in ORP. With low-risk disease, PSMs were equivalent; however, RP in the Medicare population probably has little impact on survival. This is on par with the most comprehensive review to date on this subject, by Novara et al., which reported a mean PSM rate of 15%, with a PSM rate of 9% for pT2 disease [2]. This review was primarily composed of data from high-volume academic institutions and even single-surgeon series that transitioned from a high-volume ORP surgeon to a high-volume RARP surgeon. The conclusion of this review was equivalency of PSMs by way of whatever RP approach was used, and there was a lack of data to compare survival outcomes. In contrast, in this SEER-Medicare study, a PSM difference was seen in intermediate- and high-risk disease. This is where the data become convoluted. The rates of PSMs were essentially the same for both intermediate-risk disease (15% for RARP versus 21% for ORP) and high-risk disease (15.1% for RARP versus 20.6% for ORP). How can that finding be explained, as it defies essentially all previous data? Can we believe such a study, when the PSM rate is the same for intermediate- and highrisk disease by either approach? Within 6 mo of RARP, 4.5% of patients received secondary therapy (ADT and/or RT) versus 6.2% of ORP patients, and this trend continued until 12 mo and 24 mo, respectively. There is no way of knowing whether this finding represented adjuvant or salvage therapy, since no postoperative PSA data exist. Multimodal therapy postoperatively may not be associated with worse cancer control and might even be better. There are high-level data supporting the use of adjuvant RT (refs. [37–39] in Hu et al. [1]), and perhaps the RT was delivered more appropriately in the ORP group (pT3 and/or R1). There are also data supporting the use of ADT with salvage RT (RTOG 9601; unpubl. data) and an unexpected

DOI of original article: http://dx.doi.org/10.1016/j.eururo.2014.02.015. * Corresponding author. Mayo Clinic Department of Urology, Rochester, MN 55905, USA. E-mail address: [email protected] (R.J. Karnes). 0302-2838/# 2014 European Association of Urology. Published by Elsevier B.V. All rights reserved.

674

EUROPEAN UROLOGY 66 (2014) 673–676

benefit in the control arm (SWOG S9921; unpubl. data) with adjuvant ADT for unfavorable pathology at RP. The conclusion, after all of this, is that RARP is associated with better cancer control than ORP. What is going on here? SEER-Medicare data are an invaluable resource for trends and demographics of care within the United Stated, and then their utility wanes. In general, there can be utility in using large observational data sets about treatment comparisons, particularly in the community setting, but caution and limitations abound when it comes to treatment comparison outcomes. It is a given that the comparison is not randomized, and a large randomized trial of experienced surgeons performing RARP versus ORP is unlikely to ever be conducted in Europe or North America. Plenty of observational studies have published biased results that are different than those of randomized controlled trials. A great example is when the use of ADT is looked at in primary RT men using SEER-Medicare data even after controlling for variables and matching. There is high-level evidence demonstrating a survival advantage for the use of ADT along with RT for locally advanced PCa rather than RT alone. However, when this is studied within SEER-Medicare, there is a survival disadvantage seen with the use of ADT and RT [3]. Propensity score matching, no matter how rigorously applied, may not sufficiently minimize selection bias, especially considering the inability to adjust for latent variables, patient selection, and missing data. There were obviously many missing variables in this data set, and it is known that SEER is replete with this problem and not immune to inaccurate coding— specifically, up to 30% in a study evaluating PSMs (reference 44 in Hu et al. [1]). And if one looks at the ends of the disease spectrum in a genitourinary cancer and performs an audit, inaccurate coding has been reported up to 90% of the time [4]. Higherrisk PCa features/variables have not been the norm. One cannot adjust for data one does not have (unmeasured or missing) or that are inaccurate. The number of higher-risk variables (Gleason score, cT stage, and PSA) within risk stratification matters in terms of disease pathology and outcomes [5]. We feel that cT3 is a more important variable to consider for high-risk disease, as noted by National Comprehensive Cancer Network and European Association of Urology criteria rather than cT2c. Plus, tumor volume is unaccounted for either at RP or at biopsy (number of positive cores and core percentage). Also, it is not known from this data set who had nerve sparing versus not or who might have been receiving preoperative hormonal therapy, all of which can affect the PSM rate. In addition, the number of PSMs can matter, as can the location of the PSMs in terms of survival outcomes. Since this time frame includes the earlier RARP era and its learning curve, is it not also plausible that surgeons who felt the cancer could be more locally advanced (unmeasured complexity variable), which could lead to a longer operative time, would select ORP, which was known to be ‘‘faster’’ earlier on? Perhaps surgeons felt more comfortable doing ORP for the more challenging cases. And Medicare reimbursed a flat rate to the hospital, so time mattered. Only

recently has there been a financial incentive for the surgeon to do RARP over ORP (higher relative value unit assignment). As academic centers adopted RARP, they were less likely to tackle higher-risk cancers initially with RARP; community practices were probably little different, that is, this is an unmeasured complexity variable. Emblematic of this are data from the pioneering Vattikuti team, which reported its early experience in RARP in 2007. Ten percent of the study population had PSA >10, 0.4% had cT3, and 7.7% had Gleason score 8–10. D’Amico high risk represented 8.2% of the study population [6]. Where does this leave us with the Medicare population? There is evidence that expectations are higher for RARP than for ORP (reference 25 in Hu et al. [1]), and this probably pertains to Medicare RP men as well. These men experience the same degree of functional status whether they undergo RARP or ORP [7]. In earlier experiences of ORP versus RARP (part of minimally invasive RP [MIRP]), a Medicare sample of >2700 RP men from 2003 to 2005 suggested a much higher rate of postoperative/salvage therapy for MIRP (27.8%) compared with ORP (9%) [8]. However, a different report a year later using the SEER-Medicare data from 2003 to 2007 concluded that there were similar rates of additional cancer therapies after ORP versus after MIRP (reference 2 in Hu et al. [1]). A similar study to the current one [1], on PSMs and additional cancer treatments in the period when RARP is much more common than ORP, was recently published [9] by some of these same authors on high-risk PCa. Using a SEERMedicare cohort of >1500 men undergoing RARP and ORP between 2008 and 2009, the use of additional cancer treatments and PSMs was the same. Thus, it might appear that RARP got worse with time, ORP got better with time, or both. Similar data in this ‘‘postdissemination’’ period have also shown that these men were less likely to undergo a pelvic lymph node dissection (PLND), and the men who did were more likely to have fewer nodes sampled if they underwent RARP than ORP [10]. We feel that PLND is an important part of RP in intermediate- and high-risk disease. In conclusion, we do not feel that minimum-access surgery equals maximum cancer therapy, nor are we saying that it is inferior. Let the buyer beware. Conflicts of interest: The authors have nothing to disclose.

References [1] Hu JC, Gandaglia G, Karakiewicz PI, et al. Comparative effectiveness of robot-assisted versus open radical prostatectomy cancer control. Eur Urol 2014;66:666–72. [2] Novara G, Ficarra V, Mocellin S, et al. Systematic review and metaanalysis of studies reporting oncologic outcome after robot-assisted radical prostatectomy. Eur Urol 2012;62:382–404. [3] Giordano SH, Kuo YF, Duan Z, Hortobagyi GN, Freeman J, Goodwin JS. Limits of observational data in determining outcomes from cancer therapy. Cancer 2008;112:2456–66. [4] Nguyen MM, Gill IS. Effect of renal cancer size on the prevalence of metastasis at diagnosis and mortality. J Urol 2009;181:1020–7. [5] Joniau S, Briganti A, Gontero P, et al. Stratification of high-risk prostate cancer into prognostic categories: a European multi-

EUROPEAN UROLOGY 66 (2014) 673–676

institutional study. Eur Urol. In press. http://dx.doi.org/10.1016/ j.eururo.2014.01.020. [6] Badani KK, Kaul S, Menon M. Evolution of robotic radical prostatectomy: assessment after 2766 procedures. Cancer 2007;110:1951–8.

675

[9] Gandaglia G, Abdollah F, Hu J, et al. Is robot-assisted radical prostatectomy safe in men with high-risk prostate cancer? assessment of perioperative outcomes, positive surgical margins, and use of additional cancer treatments. J Endourol 2014;28:784–91.

[7] Barry MJ, Gallagher PM, Skinner JS, Fowler Jr FJ. Adverse effects of

[10] Gandaglia G, Trinh QD, Hu JC, et al. The impact of robot-assisted

robotic-assisted laparoscopic versus open retropubic radical pros-

radical prostatectomy on the use and extent of pelvic lymph node

tatectomy among a nationwide random sample of Medicare-age

dissection in the ‘‘post-dissemination’’ period. Eur J Surg Oncol

men. J Clin Oncol 2012;30:513–8.

2014;40:1080–6.

[8] Hu JC, Wang Q, Pashos CL, Lipsitz SR, Keating NL. Utilization and outcomes of minimally invasive radical prostatectomy. J Clin Oncol

http://dx.doi.org/10.1016/j.eururo.2014.03.012

2008;26:2278–84.

Platinum Priority Reply from Authors re: R. Jeffrey Karnes, Steven Joniau, Michael L. Blute, Hendrik Van Poppel. Caveat emptor. Eur Urol 2014;66:673-5 Jim C. Hu a,*, Giorgio Gandaglia b, Karim Chamie a, Pierre I. Karakiewicz b,c, Maxine Sun b a Department of Urology, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA; b Cancer Prognostics Health Outcomes Unit, University of Montreal Health Centre, Montreal, Canada; c Department of Urology, University of Montreal Health Center, Montreal, Canada

We applaud Karnes et al. [1] for their critique of our study that demonstrates fewer positive surgical margins and less additional cancer therapy within 2 yr for robot-assisted versus open radical prostatectomy [2]. They raise concerns about the validity of our study findings. The Surveillance, Epidemiology, and End Results (SEER) program is an epidemiologic surveillance system funded by the US National Cancer Institute and consisting of population-based tumor registries that routinely collect information on all newly diagnosed cancer (incident) cases that occur in persons residing within the confines of SEER areas. The information collected about each incident cancer diagnosis includes the patient’s sociodemographic characteristics, date of diagnosis, and cancer histology, stage, and grade [3]. As mentioned in our limitations section and by Karnes et al. [1], surgical margin status for pathologic pT3b disease is not recorded by SEER, leading to these exclusions. However, men with pT3b accounted for only 6.7% of all radical prostatectomies during our study period, and 53.2% required additional cancer therapy within 24 mo of radical prostatectomy. Additionally, 59.6% versus 37.5% of pT3b subjects were classified as having high-risk versus intermediate-risk disease, and the exclusion of this significant difference contributes to the similar utilization rates of postprosta-

DOIs of original articles: http://dx.doi.org/10.1016/j.eururo.2014.02.015, http://dx.doi.org/10.1016/j.eururo.2014.03.012. * Corresponding author. University of California, Los Angeles, Urology, 924 Westwood Blvd., Los Angeles, CA 90024, USA. Tel. +1 310 405 1467. E-mail address: [email protected] (J.C. Hu).

tectomy cancer therapy for intermediate- and high-risk disease, questioned by Karnes et al. Finally, Shah et al. [4] found that 75% of pT3b had positive surgical margins, and the presence of this aggressive feature overwhelmingly prognosticates biochemical recurrence–free survival and prostate cancer–specific mortality. Thus differences in surgical margin status in the setting of seminal vesicle invasion contribute little to differences in our robotassisted versus open radical prostatectomy cancer-control outcomes. Another Hu et al. SEER-Medicare comparative effectiveness study of utilization and complications used propensity matching and found a lower likelihood of robot-assisted radical prostatectomy (RARP) transfusions, anastomotic strictures, and shorter lengths of stay, albeit with more frequently diagnosed RARP incontinence and erectile dysfunction compared with open radical prostatectomy during early adoption of RARP [5]. This finding is reinforced by worse urinary function for initial RARP versus open radical prostatectomy in a multisurgeon single-institution study that also demonstrated a shorter learning curve for continence preservation compared with open surgery that ultimately resulted in equal or better continence for RARP [6]. The Hu et al. study [5] also epitomizes caveat emptor for men considering robot-assisted surgery, and led to a defensive by surgeons performing robot-assisted procedures and the device manufacturer alike against the potential sales pitches made by open surgeons. Although SEER-Medicare data have limitations, such as the absence of postprostatectomy urinary and sexual function outcomes, the Institute of Medicine cited it as one of the few population-based data resources available for analyses of the quality of cancer care [7]. Failure of surgeons who perform robot-assisted procedures to accept increased urinary incontinence and erectile dysfunction during the learning curve or failure of an average open-procedure surgeon to acknowledge higher positive margins and adjuvant/salvage treatment rates brings to mind the timeless words of Pulitzer Prize winner Upton Sinclair, the muckraking American author: ‘‘It is difficult to get a man to understand something, when his salary depends on his not understanding it’’ [8].