In the current study, we aimed to evaluate the predictive factors of urinary leakage after PNL catheter removal following PCNL. To the best of our knowledge, this is the first report in the English literature which investigates the relation between the HU values and urinary leakage.
Since the first report of PCNL for kidney stones by Fernastom and Johanssonin in 1976 , a steadily increasing number of series have been reported in urological literature. During this period, there has been significant improvement in techniques, instruments and experience. Despite these improvements and high success rates, PCNL still has many complications such as blood loss, urinoma, urosepsis and adjacent organ injury . UL is also a major postoperative complication after PNL catheter removal with a reported incidence up to 70% [3, 4]. In another study, the prevalence of stent requirement after PCNL was reported as 4.3–5% . In our experience, the overall incidence of UL was 40.1% and 11(2.7%) of our patients needed stent replacement.
The literature about the BMI effect on PCNL outcomes is unclear. Some reports concluded that BMI had an impact on complication rates of PCNL [11, 12], while some studies noted that BMI did not affect the UL status after PCNL [13, 14]. Dirim et al.  noted that BMI did not affect the UL status after PCNL. Our findings are consistent with this report.
To the best of our knowledge, the effect of diabetes mellitus on urinary leakage has not been evaluated before. In the present study, we also found that the presence of DM increased the urinary leakage incidence. As we know, there are many factors that contribute to the altered tissue repair of DM . DM causes microvascular disease which results decreased blood flow and insufficient oxygen delivery. These factors may delay the closure of renal parachimal puncture which may cause urinary leakage. On multivariate analysis, DM was found to be an independent predictive factor for UL after PNL catheter removal.
Dirim et al.  and Ansari et al.  did not demonstrate correlation between the stone burden and UL. However, as opposed these results, we found that UL was correlated with both stone surface area and stone burden. Additionally, we compared the stone localization and found that multiple calyceal and staghorn stones were also correlated with UL. We thought that all these stone characteristics make surgery more difficult and causes more complications.
Previous studies evaluated the relationship between the hydronephrosis and UL [3, 4, 10]. All these studies concluded that UL increases with the degree of HN. In the present study, we also confirmed the same results. Furthermore, due to our subgroup analyses, UL is statistically significant in high-grade HN (grade 2–3).
The association between renal PT and UL has been investigated in some studies. Uyetürk et al.  found 17.2 mm optimum cut-off value of PT for hospitalization ≤ 12 h with 90.2% sensitivity and 69.4% specificity. Ansari et al.  determined 17 mm cut-off value of PT hospitalization of patients due to prolonged urinary leakage with 95.2 sensitivity and 60.2% specificity. The present study also confirmed that the PT in access line is inversely correlated with UL on multivariate analyses.
The HU values determined in NCCT provide information about stone density and stone formation. There are a lot of publications about HU values and outcomes of the treatment option like SWL and PNL. Quzaid et al.  reported that HU threshold of 970 was predictive for successful ESWL. While Gok et al.  did not find any correlation between HU values and PNL success rates, Gücük et al.  reported higher success rates with lower HU values. Although the correlation between the HU values and success rates has been studied many times until now, the relationship between the HU values and UL status following PCNL procedure has not been discussed in literature so far. In the present study, we hypothesized that the higher HU values would be correlated with UL and multivariate analyses confirmed this hypothesis. We tried to obtain a threshold for HU values. There are no published data on this issue. When we performed the ROC curve analyses, the optimum cut-off value of HU for UL was 933 with 84.9% sensitivity and 67.1% specificity.
Dirim et al.  had pointed out that the presence of previous renal stone surgery or SWL treatment had no impact on urinary leakage following PCNL. In subgroup analysis, this study also demonstrates that previous SWL did not affect leakage status, whereas our findings indicate that open surgery or PCNL history had effect on urinary leakage. But on multivariate analyses none of them remained significant.
Differently from the literature, we also evaluated the effect of surgery related factors incluiding operation and fluoroscopy time, treatment outcome, j stent use, PCN catheter stay time and hospitalization time on UL. On univariate analysis we observed that all these factors had effect on urinary leakage. When we performed multivariate analysis, fluoroscopy time, j stent use and PCN catheter stay time were found to be independently and significantly associated with UL.
Our study is not without limitations. First, it has a retrospective design. As it is the first study which investigates the effect of DM presence on UL, we could not reach Hba1c levels for all patients. We only reached the blood glucose levels and arterial tension values during hospitalization time by the clinical progresses.