Research Correspondence: Extended pelvic lymph‐node dissection is independently associated with improved overall survival in patients with PCa at high‐risk of lymph‐node invasion

Dear Editor,

It is generally agreed upon that an extended pelvic lymph‐node dissection (ePLND) provides valuable staging information and helps guide adjuvant therapy, and thus should be undertaken in prostate cancer patients with aggressive preoperative disease features at the time of radical prostatectomy [1,2]. However, whether it has a ‘direct’ therapeutic benefit in the aforesaid patients has remained difficult to demonstrate [3]. The only patients that seem to derive a survival advantage from an ePLND are patients with pN1 disease [4] – this cited study suggested a direct therapeutic effect of an ePLND, with a 7% incremental benefit in 10‐year cancer‐specific survival per every additional LN removed (P  = 0.02). However, it did not identify these patients preoperatively.

Given the significant side‐effects associated with an ePLND [3], it is worth asking the questions: which patients, identified preoperatively, may derive a direct therapeutic benefit from an ePLND, and who benefit indirectly only (i.e. via optimal utilisation of adjuvant therapies). The latter question has been answered [5,6]. Here, we try to answer the former.

We relied on the National Cancer Database (NCDB) to answer our question. The NCDB, a joint programme of the Commission on Cancer and the American Cancer Society, is a nationwide cancer database that contains information on ~70% of newly diagnosed tumours in the USA. We identified all patients with prostate cancer undergoing radical prostatectomy between the years 2004 and 2015. After excluding patients with clinical LN/metastatic disease (n  = 2568), neoadjuvant radiotherapy, chemotherapy or hormonal therapy (n  = 10 931), missing information on biopsy Gleason score, cT stage or preoperative PSA value (n  = 166 696), and missing information regarding PLND (n  = 95 348), a final sample of 311 061 patients was achieved. All available baseline patient/tumour characteristics and overall survival (OS) data (outcome) were noted. Preoperative LN invasion (LNI) risk was calculated using the Godoy nomogram. We used this nomogram as it was developed using the PLND data from North American men, and has been validated in them [6]. The cut‐off of ≥10 LNs to define an ePLND was based on prior studies [5,6,7,8]. To analyse the impact of ePLND (≥10 LNs) vs none/limited PLND (0–9 LNs) on 10‐year OS, interaction between Godoy nomogram predicted LNI probability, which is based on the preoperative PSA value, clinical stage and biopsy Gleason grade, and ePLND/PLND was plotted using locally weighted methods controlling for age, comorbidities and adjuvant radiation therapy (aRT). This was called model 1 (M1). In a second model (M2), in addition to controlling for age, comorbidities and aRT, we also adjusted for receipt of adjuvant hormonal therapy (aHT). We performed this analysis as we reasoned that a survival benefit in patients undergoing an ePLND may be due to better staging and receipt of aHT. All analyses were performed with the Statistical Analysis System (SAS), version 9.4 (SAS Institute, Cary, NC, USA), with a two‐sided P  < 0.05 considered as statistically significant. An Institutional Review Board waiver was obtained prior to conducting this study, in accordance with institutional regulations on dealing with de‐identified administrative data.

Table S1 provides baseline characteristics. Of the 311 061 patients, 49 470 (15.9%) patients underwent an ePLND. The median number of LNs removed in patients undergoing none/limited PLND vs ePLND were 2 and 14, respectively (P  < 0.001). The median age and preoperative PSA values for the groups were 61 and 62 years (P  < 0.001) and 5.5 and 6 ng/mL (P  < 0.001), respectively. Patients undergoing an ePLND had more aggressive disease on pathological analysis: Gleason ≥8 disease (17.3% vs 10.0%), pT3+ stage (37.4% vs 21.9%) and pN1 disease (8.6% vs 1.5%; P  < 0.001 for all). These patients also received aRT (3.9% vs 3.1%) and aHT (4.3% vs 1.9%) more frequently than patients undergoing none/limited PLND (P  < 0.001 for both).

The median (interquartile range) follow‐up for the ePLND and none/limited PLND groups was 54.0 (31.3–79.9) and 57.5 (35.1–82.0) months, respectively. In interaction analyses, the lines for ePLND and none/limited PLND separated at Godoy nomogram predicted LNI risk of 20% in model M1 (Fig. 1a), indicating that patients with a preoperative LNI risk >20% derived an OS benefit from an ePLND. This finding remained preserved in model M2, which adjusted for receipt of aHT, in addition to age, comorbidities and aRT, thus indicating a ‘direct’ independent benefit of an ePLND on OS in patients with a LNI risk of >20% (Fig. 1b).

In Cox regression analyses, the first model (M1) demonstrated that patients undergoing an ePLND (hazard ratio [HR] 1.20, 95% CI 1.17–1.24) had a 9% incrementally lower hazard of 10‐year mortality than patients undergoing none/limited PLND (HR 1.29, 95% CI 1.26–1.31) for every 10% increment in Godoy nomogram predicted LNI risk, beyond the 20% cut‐off (P  < 0.001). Similarly, the second model (M2) demonstrated that patients undergoing an ePLND (HR 1.18, 95% CI 1.14–1.21) had a 6% incrementally lower hazard of 10‐year mortality than patients undergoing none/limited PLND (HR 1.24, 95% CI 1.23–1.26) for every 10% increment in Godoy nomogram predicted LNI risk, beyond the 20% cut‐off (P  < 0.001). This lower but preserved incremental improvement in OS after adjustment for aHT (model M2) supports our hypothesis that an ePLND is in itself a ‘direct’ independent factor in OS in patients at high‐risk of LNI.

The current American and European urological societal guidelines recommend performing an ePLND in high‐risk and unfavourable intermediate‐risk patients, especially when the estimated risk for LNI is >5% [1, 2]. However, at this cut‐off, the benefit is mainly that of accurate staging and subsequent optimal adjuvant treatment (indirect benefit). This must be balanced against the morbidity of an ePLND. In line with this, a recent exhaustive systematic review by Fossati et al. [3] found that ePLND, as it is currently utilised, is associated with increased risk of postoperative complications without an oncological benefit. The findings of our present study are thus timely and important. We for the first time identify patients preoperatively that may derive both direct and indirect therapeutic benefits of an ePLND. In the present study 4.5% of the 311 061 patients had a LNI risk of >20%. This constitutes a substantial number of patients. These patients should be strongly advised to receive an ePLND. For patients constituting the LNI risk group between 5% and 20%, they should still be encouraged to undergo an ePLND after discussing the risks and benefits of it, as accurate staging may improve their survival by receipt of aHT.

Our present study is not devoid of limitations. First, it is limited by its retrospective nature, an inherent drawback of all observational studies based on administrative data. Therefore, our findings should be interpreted with caution. However, randomised data on this subject are currently scarce. The two randomised trials (NCT01812902 and NCT01555086) comparing ePLND vs limited PLND have not yet matured to provide clinically meaningful information. While we await results from these trials, our present study provides an avenue to have an informed discussion with the patients with high‐risk prostate cancer about the risks/benefits of undergoing an ePLND. Second, no centralised pathological review was available in our study. While this might be considered a limitation, it is also a strength, as it implies that our results are applicable to clinical practice, regardless of pathology review variation. Lastly, the definition of our ePLND was based on number of LNs removed rather than the anatomical zones dissected [7]. The information regarding LN zonal anatomy is not available within NCDB; however, several prior studies of anatomical ePLND have shown median LN counts between 10 and 20 [5,6,7,8], and it was 14 in our series for patients undergoing an ePLND (vs a median of two LNs for none/limited PLND), thus suggesting that the patients were likely classified appropriately into ePLND and none/limited PLND groups.

Limitations notwithstanding, our present study is the first to preoperatively identify patients in whom an ePLND may confer a direct survival advantage, in addition to superior prognostication (indirect benefit). As we identify these patients preoperatively, this may facilitate patient counselling and optimal utilisation of ePLND.

References

1. Sanda MG, Cadeddu JA, Kirkby E et al. Clinically localized prostate cancer: AUA/ASTRO/SUO guideline. Part II: recommended approaches and details of specific care options. J Urol 2018; 199: 990– 7
2. Mottet N, Bellmunt J, Bolla M et al. Guidelines on prostate cancer. Part 1: screening, diagnosis, and local treatment with curative intent. Eur Urol 2017; 71: 618– 29
3. Fossati N, Willemse PM, Van den Broeck T et al. The benefits and harms of different extents of lymph node dissection during radical prostatectomy for prostate cancer: a systematic review. Eur Urol 2017; 72: 84– 109
4. Abdollah F, Gandaglia G, Suardi N et al. More extensive pelvic lymph node dissection improves survival in patients with node‐positive prostate cancer. Eur Urol 2015; 67: 212– 9
5. Briganti A, Larcher A, Abdollah F et al. Updated nomogram predicting lymph node invasion in patients with prostate cancer undergoing extended pelvic lymph node dissection: the essential importance of percentage of positive cores. Eur Urol 2012; 61: 480–7
6. Godoy G, Chong KT, Cronin A et al. Extent of pelvic lymph node dissection and the impact of standard template dissection on nomogram prediction of lymph node involvement. Eur Urol 2011; 60: 195– 201
7. Weingartner K, Ramaswamy A, Bittinger A, Gerharz EW, Voge D, Riedmiller H. Anatomical basis for pelvic lymphadenectomy in prostate cancer: results of an autopsy study and implications for the clinic. J Urol 1996; 156: 1969– 71
8. Abdollah F, Sun M, Thuret R et al. Lymph node count threshold for optimal pelvic lymph node staging in prostate cancer. Int J Urol 2012; 19: 645– 51