Patient Factors Impacting Perioperative Outcomes for T1b-T2 Localized Renal Cell Carcinoma May Guide Decision for Partial versus Radical Nephrectomy

There remains debate surrounding partial (PN) versus radical nephrectomy (RN) for T1b-T2 renal cell carcinoma (RCC). PN offers nephron-sparing benefits but involves increased perioperative complications. RN putatively maximizes oncologic benefit with complex tumors. We analyzed newly available nephrectomy-specific NSQIP data to elucidate predictors of perioperative outcomes in localized T1b-T2 RCC. We identified 2094 patients undergoing nephrectomy between 2019–2020. Captured variables include surgical procedure and approach, staging, comorbidities, prophylaxis, peri-operative complications, reoperations, and readmissions. 816 patients received PN while 1278 received RN. Reoperation rates were comparable; however, PN patients more commonly experienced 30-day readmissions (7.0% vs. 4.7%, p = 0.026), bleeds (9.19% vs. 5.56%, p = 0.001), renal failure requiring dialysis (1.23% vs. 0.31%, p = 0.013) and urine leak or fistulae (1.10% vs. 0.31%, p = 0.025). Infectious, pulmonary, cardiac, and venothromboembolic event rates were comparable. Robotic surgery reduced occurrence of various complications, readmissions, and reoperations. PN remained predictive of all four complications upon multivariable adjustment. Several comorbidities were predictive of complications including bleeds and readmissions. This population-based cohort explicates perioperative outcomes following nephrectomy for pT1b-T2 RCC. Significant associations between PN, patient-specific factors, and complications were identified. Risk stratification may inform management to improve post-operative quality of life (QOL) and RCC outcomes.


Introduction
Renal cell carcinoma (RCC) is associated with over 70,000 diagnoses and 14,000 deaths annually in the United States [1][2][3]. Incidence increased by 3-4% annually over the past 25 years, partly due to stage migration from improved diagnostic imaging [4]. Although radical nephrectomy (RN) was previously standard for all renal tumors, earlier detection and improving robotic technologies have precipitated growing utilization of partial nephrectomy (PN) [5][6][7].
Nephron-sparing surgery offers long-term benefits including reduced chronic kidney disease (CKD) and associated cardiovascular or metabolic sequelae [8][9][10]. Conversely, retrospective studies demonstrate increased perioperative complications with PN, often attributed to its technical complexity [1,11]. Importantly, these results may be confounded by institutional variations in surgical technique, demographics, and selection bias. Accordingly, the literature is largely inconclusive; a separate analysis of cT2 masses found comparable perioperative complications and survival [11]. European Organisation for Research and Treatment of Cancer trial 30904 demonstrated oncologic equivalence, though perioperative complications including bleeding, urinary fistulae, and reoperations were increased with PN; however, enrollment concluded in 2003 and techniques have since improved, particularly with the advent of minimally invasive surgery (MIS) [12][13][14].
Presently, PN is recommended for T1a RCC, while RN remains customary for locally advanced T3 tumors [15,16]. However, there remains significant equipoise for T1b-T2 disease, where decisions are largely based on gestalt. The National Comprehensive Cancer Network recommends either PN or RN, while various associations offer conflicting practice guidelines [17,18]. Three scores, the RENAL nephrometry score, PADUA, and c-index are commonly used to facilitate decision-making based on anatomic factors, but these systems do not account for non-anatomic patient-specific factors such as comorbidities.
Patient-and tumor-specific characteristics, including demographics, comorbidities, tumor size, and histopathology, were previously demonstrated to impact survival following PN versus RN for T2 masses [6,8]. Although risk factors affecting long-term outcomes have been compared, their predictive value for near-term complications remains unclear. Ultimately, clinicians must consider whether increased perioperative morbidity offsets the advantages of PN, and whether appreciating risk factors can allow prediction and mitigation of such morbidity. Considering the equipoise in this disease space, updated guidance would improve localized RCC management [13,19].
The National Surgical Quality Improvement Program (NSQIP) database captures standardized data on baseline characteristics and perioperative outcomes from over 600 hospitals [20]. By utilizing NSQIP and its new Nephrectomy Participant Use Data File (PUF), this study captures a modern, cross-institutional cohort to elucidate nephrectomy outcomes and predictors of perioperative morbidity. We hypothesized that patient-specific factors associated with specific perioperative outcomes would guide PN versus RN decision-making for T1b-T2 RCC in cases of oncologic equivalence.

Description of Covariates
Variables of interest included age, race, ethnicity, pathologic stage, and pre-operative diabetes, smoking, respiratory conditions, cardiovascular diseases, chronic steroid use, bleeding disorders, and creatinine. Complications of interest included surgical site infections (SSIs); urinary tract infections (UTIs); sepsis; wound dehiscence; cardiovascular, pulmonary, and renal events; venothromboembolism (VTE); bleeding events within 72 hours requiring transfusion; and urine or lymphatic leaks, fistulae, or obstructions. NSQIP acute renal failure, defined as new post-operative dialysis requirement, was denoted "renal failure requiring dialysis", while NSQIP progressive renal insufficiency, defined as creatinine increase >2mg/dL without dialysis, was denoted "acute kidney injury (AKI)". Surgical variables including stent and drain placement, perioperative antibiotics, VTE prophylaxis, and surgical approach were described, as were quality measures including operative time, length of hospital stay (LOS), reoperations within 30 days, and readmissions within 30 days.

Statistical Analysis
Descriptive statistics for demographics and complications were compared by nephrectomy type, surgical approach, and comorbidities using Fisher's exact test and Pearson's chi-square test. Multivariable logistic regression hazard models were generated to report odds ratios (ORs) comparing complication rates between PN and RN following correction for associations (demographics, comorbidities, pre-operative measurements, and surgical approach) deemed significant on univariate analysis or clinically relevant a priori.
All statistical tests were two-tailed. p < 0.05 was considered statistically significant. Statistical analyses were performed using GraphPad Prism 9.0.2 (San Diego, CA, USA) and SPSS 28.0.1 (IBM Corp., Armonk, NY, USA).

Impact of Surgical Approach
Given the expanding indications of MIS, complication rates were also compared by surgical approach (Supplementary Table S1

Predictors of Perioperative Events
Univariate analysis uncovered comorbidities associated with particular complications (Table 3). Upon multivariable analysis (MVA), with nephrectomy type, approach, and relevant comorbidities included as covariates, patient-specific predictors were uncovered (Table 4). MVA was repeated for the PN and RN cohorts separately to further clarify risk factors (Supplementary Tables S2 and S3).

Infectious
SSIs and sepsis events were not predicted by any patient factors. Wound dehiscence was associated with pre-operative renal failure (OR 175.494, p < 0.001). Notably, pneumonia was predicted by COPD (OR 5.789, p < 0.001) and bleeding disorder (OR 6.596, p < 0.001). The occurrence of UTIs was only predicted by hypertension (OR 3.012, p = 0.047). Nephrectomy type did not influence infectious events.

Cardiovascular
Although diabetes was predictive of MI (OR 4.988, p = 0.013), no patient factors were predictive when cardiovascular events-including MI, CVA, and cardiac arrest-were combined as a single outcome. Nephrectomy type did not predict any cardiovascular events.

Renal
Renal failure requiring dialysis was more common following PN, although this lost significance as a predictor following multivariable adjustment for surgical approach (OR 2.811, p = 0.066).
Pre-operative acute renal failure was a strong predictor within the PN cohort (OR 36.961, p = 0.016). No pre-operative renal failure patients receiving RN experienced this complication.
Similarly, open surgery was highly predictive within the PN cohort (OR 7.323, p = 0.014), but not within the RN cohort (OR 3.598, p = 0.206).
Hypertension predicted AKI on MVA (OR 7.689, p = 0.049). After covariate adjustment, PN patients appeared slightly more prone to experience AKI (OR 2.740, p = 0.098), although this did not reach statistical significance.

Urine and Lymph Leak
Although urine leaks or fistulae occurred more commonly in PN patients, this was not predicted by nephrectomy type following multivariable adjustment for surgical approach (OR 3.000, p = 0.073). Open surgery did not demonstrate different effects with PN (OR 2.482, p = 0.182) versus RN (OR 4.067, p = 0.164). Ureteral obstructions were not associated with any risk factors.
Dialysis (OR 6.680, p = 0.005) and renal failure (OR 12.369, p = 0.040) were predictive of lymphatic leaks on MVA, while procedure type was not.
Reoperations were predicted by pre-operative renal failure, dialysis, steroid use, and bleeding disorders on MVA, while nephrectomy technique did not affect reoperation rates.

Discussion
This study analyzed a representative population of 2094 localized T1b/T2 RCC patients receiving nephrectomy from 2019-2020. The NSQIP Procedure-Targeted PUFs have never been previously utilized to describe urology-specific quality metrics, and the literature lacks a population-based comparison of the competing short-term risks and benefits of PN and RN [11]. Previous studies of smaller cohorts were limited in generalizability or stratification by case-specific factors.
Patients with pre-existing diabetes, hypertension, or cardiovascular risk often undergo nephron preservation. Associated benefits should be weighed with risks of perioperative complications following this technically complex procedure. Understanding patient-specific factors modulating that risk may facilitate treatment choice.
Demographic differences between our cohorts, including increased PN utilization in younger males with lower tumor stages, are supported by previous studies [6,9]. The disparity in operative time, which is well-described, serves as a surrogate measure demonstrating the increased technical complexity of PN. Differences in perioperative morbidity are also supported [12,14,21]. However, no studies have identified specific patient populations in which perioperative complications were actually relevant, hence precluding real-world application. A study of 2011 NSQIP data found that readmissions following RN were not predicted by demographic factors, but comorbidity data were not analyzed [22].
Considering the decrease in nephrectomy complication rates over time, along with recent improvements in MIS, an updated analysis is required [5]. Robotic PN demonstrates reduced complications, bleeding, and readmissions than open PN [5,23,24]. Our modern cohort corroborated this trend, demonstrating that increased adoption of MIS may expand indications for PN while mitigating its associated peri-operative risks.
This study suggests significant associations between PN and the following four complications: bleeding, renal failure requiring dialysis, ureteric leaks or fistulae, and readmissions. This understanding may either inform preventative measures to reduce risk for PN patients or guide surgeons towards RN in cases where these complications would be undesirable. Notably, within the PN cohort, MIS was associated with fewer cases of renal failure requiring dialysis and urine leaks. Certainly, MIS was generally safer and more common with PN than RN, but it is possible that this association also reflects surgeon selection bias. Notably, adjustment for any patient-specific comorbidities lacked the same effect, indicating that these two complications are still important considerations, as patients with anatomically complex tumors are likely to continue to receive open surgery.
Ultimately, it appears that although PN offers several benefits for localized RCC patients and has low complication rates, those with certain baseline characteristics may fare better with RN. Clinician awareness of differential risks following PN versus RN can facilitate patient-specific treatment choice based on values surrounding post-operative QOL. For instance, patients in whom bleeds would be particularly undesirable (e.g., bleeding disorder or Jehovah's witness patients) may be poorer candidates for PN. Those interested in improved short-term QOL may prefer RN to avoid readmission.
Although renal failure requiring dialysis was significantly increased for PN patients, even after multivariable correction for baseline renal function, this may be attributable to selection bias. Patients with low preoperative renal function or CKD are frequently recommended PN considering their higher risk for end stage renal disease (ESRD) progression. The NSQIP database does not include pre-operative CKD, and the available pre-operative renal failure variable only captures acute declines in renal function within 24 h prior to the index operation [16,25]. Our PN cohort may contain a disproportionate amount of baseline CKD patients. Recent findings suggest that such patients may instead benefit from active surveillance or upfront RN with dialysis or kidney transplantation listing to avoid renal and extrarenal morbidities following PN [25]. While post-PN progression to ESRD is uncommon in the general population, patients with preexisting low renal function are at risk, with studies estimating an occurrence of over 15%, increasing with tumor size and complexity as nephron preservation becomes limited [25].
Initial univariate analysis revealed several additional comorbidities associated with other perioperative complications which were not different between PN versus RN. Understanding these trends, though not controllable by changing nephrectomy type, is still important, as urologists can anticipate and manage complications. Hypertension was previously demonstrated to influence renal functional outcomes after PN [26]. For example, increased DVT rates with chronic steroid use or PE risk from COPD may suggest need for enhanced VTE prophylaxis in select groups. Similarly, increased MI, bleeds, and UTIs in patients with diabetes may endorse a need for improved cardiac monitoring, antibiotic prophylaxis, and wound closure strategies. Procedure-specific risk calculation indices have been previously proposed, including an algorithm predicting risk of major adverse cardiovascular events following PN based on demographics and baseline factors [27]. Interestingly, this algorithm does not include diabetes, although our study identified diabetes as a significant predictor of MI. Therefore, our analysis indicates the importance of further research to update existing risk calculation algorithms or create new ones where applicable.
Although NSQIP is a high-quality database drawing data from trained institutional reporters, this study has limitations [28,29]. Foremost, its retrospective nature limits interpretation. Moreover, institutional participation in NSQIP is voluntary, and case patterns may not represent all nephrectomies being performed. This study includes data over a one-year period (2019-2020), and findings including surgeon decision-making or patient demographics may have been impacted by the height of the COVID-19 pandemic. Lastly, although 30-day outcomes are a key metric of surgical safety, treatment success also involves long-term outcomes including cancer recurrence and survival, which are not captured by NSQIP. Similarly, imaging results or tumor-specific variables such as nearness to collecting system and overall anatomy, as assessed by RENAL nephrometry scores, are important considerations in surgical assignment, but are not available in NSQIP [7,30,31]. Hence, the findings of this study may only be applied in cases of oncologic equivalence.

Conclusions
This contemporary analysis demonstrates the value of incorporating case-specific factors within an algorithm to inform T1b-T2 RCC management and patient counseling. Complication rates between PN and RN differed significantly, with higher rates of bleeding, urine leak or ureteric fistulae, renal failure requiring dialysis, and readmissions following PN. However, both techniques remained extremely safe, particularly for patients lacking described risk factors. These findings suggest that increasing PN usage is well-advised, although certain scenarios may warrant risk-based management or RN use.  Table S2. Multivariate analysis of associations between patient factors and complications of interest following partial nephrectomy, Supplemental Table S3. Multivariate analysis of associations between patient factors and complications of interest following radical nephrectomy.