Percutaneous cryoablation of renal masses: Washington University experience of treating 129 tumours

Authors


Abstract

What's known on the subject? and What does the study add?

  • For patients who are unfit for extirpative surgery, percutaneous cryoablation (PCA) presents a minimally-invasive alternative for the treatment of renal masses. PCA has been demonstrated to be safe, with complication rates <10% being reported consistently. Studies have suggested that a minimal and insignificant decline in renal function can occur after PCA. Finally, among studies with a follow-up >20 months, treatment success rates range from 75% to 96%. However, longer-term oncological and functional results for patients treated with PCA are relatively limited.
  • The present study profiles one of the largest reported experiences with PCA for renal masses: 129 tumours in 124 patients. Our complication rate was comparable to that observed in other reported studies. At a mean follow-up of 30 months, treatment success was achieved in 87% of tumours, which is in line with published PCA success rates. On multivariable analysis, tumour size >3.0 cm was found to be significantly associated with treatment failure. A minimal but statistically significant renal functional decline was observed, with 20% of patients experiencing a progression in National Kidney Foundation-Chronic Kidney Disease stage. On multivariable analysis, age >70 years, hilar tumour location and postoperative day 1 estimated glomerular filtration rate <60 mL/min/1.73 m2 were found to be significantly associated with renal functional decline. The present study confirms that PCA of renal masses represents a safe alternative to surgery in patients with substantial medical comorbidities. In the present cohort, baseline patient and tumour characteristics probably impact the risk of tumour recurrence, as well as renal disease progression, after PCA.

Objective

  • To evaluate perioperative, oncological and functional outcomes after percutaneous cryoablation (PCA) for renal masses based on our single-centre experience.

Patients and Methods

  • We retrospectively identified 124 patients who underwent PCA for 129 renal tumours between March 2005 and June 2011.
  • Patient demographics and baseline clinical characteristics, tumour features, perioperative information, and postoperative outcomes were recorded.
  • Oncological outcomes were defined by radiographic evidence of recurrence on follow-up computed tomography or magnetic resonance imaging.
  • Renal disease progression was defined by a change in National Kidney Foundation-Chronic Kidney Disease stage.

Results

  • Patients had mean (sd) age of 72.6 (10.2) years; mean (sd) tumour size and nephrometry score were 2.7 (1.1) cm and 6.5 (1.7), respectively.
  • Our overall complication rate was 9% (11/124), whereas the major (greater than Clavien II) complication rate was 2% (2/124).
  • Significant predictors of renal disease progression following PCA included age ≥ 70 years (odds ratio [OR], 4.31, P = 0.03), hilar tumour location (OR, 4.67, P = 0.04), and post operative day 1 estimated glomerular filteration rate ≤60 mL/min/1.73 m2 (OR, 7.09, P = 0.02).
  • Our treatment success rate was 87% (112/129) at a mean (sd) follow-up of 30.2 (18.8) months.
  • Tumour size ≥3.0 cm was significantly associated with PCA failure (hazard ratio, 3.21, P = 0.03).

Conclusion

  • PCA provides a safe and oncologically effective alternative to extirpative surgery for renal masses in patients with significant medical comorbidities.

Introduction

As a result of the widespread utilization of diagnostic abdominal imaging, the incidence of asymptomatic small renal masses (SRMs) has continued to increase [1]. In the USA, the estimated number of new kidney cancer cases has increased from 39 000 in 2006 to 61 000 in 2011 [2, 3]. Over the last decade, nephron-sparing surgery has replaced radical nephrectomy as the preferred treatment option for SRMs [4-6]. However, the greatest incidence of SRMs occurs in elderly patients, who may have multiple comorbidities that preclude extirpative surgery [7]. In these patients, laparoscopic cryoablation (LCA) and image-guided percutaneous cryoablation (PCA) have shown decreased morbidity, as well as short-term oncological success [8-11]. Furthermore, recent studies suggest that PCA may reduce hospital stay and lower complication rates compared to LCA [12-14]. The present retrospective study aimed to evaluate perioperative, oncological and functional outcomes after PCA for renal masses based on our single-centre experience.

Patients and Methods

With Institutional Review Board approval, we performed a retrospective review of patients who underwent PCA for renal masses at our institution between March 2005 and June 2011. A total of 124 patients who underwent PCA were identified and included in the analysis. All patients underwent CT or MRI that showed an enhancing mass suspicious for RCC before surgery. PCA was performed on a total of 129 tumours in 124 patients. Indications for PCA included medical comorbidity, previous nephrectomy or otherwise compromised baseline renal function, multiple renal masses, and patient and surgeon preference.

PCA Technique

After CT or MRI review, PCA was performed by the Vascular and Interventional Radiology Service. Routine preoperative laboratory tests were conducted. PCA was performed under real-time CT guidance. Conscious sedation was utilized in almost all cases, with the exception of patients with a low probability of tolerating extended prone positioning. Those patients had PCA performed under general anaesthesia. Because of the risk of haemorrhage and the associated negative impact on tumour visualization, biopsies were not routinely performed. After routine skin preparation, sterile draping and application of local anaesthetic to the insertion sites, cryoprobes were advanced percutaneously with the patient remaining still after deep inspiration. Cryoprobes were advanced to just beyond the deep margin of the tumour, and placement was confirmed by CT. Depending on the tumour size, single or multiple cryoprobes were utilized to ensure adequate ice ball coverage. Generally, three probes were used for each tumour treated. Our initial eight patients were treated with 3.4-mm cryoprobes (Oncura Inc., Plymouth Meeting, PA, USA); the subsequent 116 patients were treated with 1.7- and 2.4-mm cryoprobes (Endocare Inc., Irvine, CA, USA). The initial freeze cycle was performed for 10 min, at temperatures below −40 °C. CT images were obtained at 5 and 10 min. After an 8-min passive thaw period, cryoablation was repeated for an additional 10 min. During the second freeze cycle, CT images were repeated at 5 and 10 min, and appropriate ice ball formation was confirmed. After at least 5 min of active thaw, the cryoprobes were removed once temperatures were above 20 °C. Contrast-enhanced imaging (CT or MRI) and serum creatinine were obtained at 1, 3, 6 and 12 months, and yearly thereafter.

Data Collection and Analysis

Patient demographics, tumour characteristics, perioperative information, follow-up imaging, and pre- and postoperative serum creatinine were recorded. Baseline comorbidities were assessed using the Charlson comorbidity index (CCI) [15]. Tumours were classified according to the nephrometry score [16]. Oncological outcomes were determined by radiographic evidence of tumour recurrence, which was defined by contrast enhancement at the cryoablation site on follow-up imaging. Estimated glomerular filtration rate (eGFR) was calculated before PCA and at most recent follow-up using the Modification of Diet in Renal Disease equation [17]. Renal disease progression was considered to have occurred if the patient experienced a deterioration in National Kidney Foundation-Chronic Kidney Disease (NKF-CKD) stage during follow-up [18]. Postoperative complications were recorded and classified by Clavien–Dindo grade [19]. Statistical analysis was performed using SPSS, version 20.0 (IBM, Armonk, NY, USA). P < 0.05 (two-tailed) was considered statistically significant. For the multivariate analyses, variables were included in the statistical models based on presumed clinical correlation or if P ≤ 0.10 was noted on Fisher's exact test.

Results

Patient characteristics are summarized in Table 1 total of 124 patients with a mean (sd) age of 72.6 (10.2) years underwent PCA for 129 tumours. Mean body mass index was 32.1 (9.2) kg/m2. Mean (sd) CCI and age-adjusted CCI were 3.9 (1.8) and 6.4 (2.0), respectively. Mean (sd) preoperative eGFR was 70.3 (24.5) mL/min/1.73 m2. Tumour characteristics are summarized in Table 2. Mean (sd) tumour size was 2.7 (1.1) cm and mean (sd) nephrometry score was 6.5 (1.7).

Table 1. Patient characteristics.
CharacteristicValue
  1. CCI, Charlson comorbidity index; eGFR, estimated glomerular filtration rate.
Age (years), mean (sd)72.6 (10.2)
Body mass index (kg/m2), mean (sd)32.1 (9.2)
Participants, n (%)76/124 (61)
Ethnicity, n (%) 
Caucasian102/124 (82)
African-American15/124 (12)
Other7/124 (6)
CCI, mean (%)3.9 (1.8)
Age-adjusted CCI, mean (%)6.4 (2.0)
Mean preoperative eGFR (mL/min/1.73 m2), mean (%)70.3 (24.5)
Table 2. Tumour characteristics.
CharacteristicValue
Nephrometry score, mean (sd)6.5 (1.7)
Tumour size (cm), mean (sd)2.7 (1.1)
Tumours ≥3 cm, n (%)44 (34%)
Right-sided, n (%)63 (49%)
Tumour depth, n (%) 
≥50% exophytic63/129 (49)
<50% exophytic52/129 (40)
Completely endophytic14/129 (11)
Mean distance to collecting system (mm), mean (sd)7.0 (4.6)
Tumours with distance ≤4 mm, n (%)32/129 (25)
Anterior–posterior location, n (%) 
Anterior23/129 (18)
Lateral21/129 (16)
Posterior85/129 (66)
Polar location, n (%) 
Upper pole34/129 (26)
Mid-zone38/129 (29)
Lower pole56/129 (43)
Hilar14/129 (11)
Tumour stage, n (%) 
T1a tumour114/129 (88)
T1b tumour15/129 (12)

Perioperative and postoperative information is summarized in Table 3. Mean (sd) procedure time was 186 (40) min. There were eight intra-operative (6%) and three postoperative complications (2%). Intra-operative complications consisted of haematoma formation that was treated conservatively (n = 6) and oxygen requirement as a result of desaturation (n = 2). Postoperative complications consisted of blood transfusion in the setting of large haematoma and decreased haematocrit (n = 2) and a duodenal and ureteral injury that resulted in a persistent retroperitoneal abscess and duodenal-ureteral fistula (n = 1). The duodenal and ureteral injury occurred in a patient with a 4.3-cm, right-sided tumour, nephrometry score 8a. This presumed intra-operative injury was diagnosed on postoperative day 28, when the patient presented with abdominal pain and a CT scan was performed. At that time, an abscess drain and a nephrostomy tube were placed, and the complication was managed non-operatively. At present, all drains have been removed.

Table 3. Perioperative and postoperative outcomes.
VariableValue
  1. eGFR, estimated glomerular filtration rate. *Renal disease progression was defined according to the National Kidney Foundation-Chronic Kidney Disease classification system [18].
Procedure time (min), mean (sd)186 (40)
Length of stay (days), mean (sd)1.3 (0.7)
Intra-operative complications, n (%)8/124 (6)
Postoperative complications, n (%)3/124 (2)
Clavien I0 (0)
Clavien II2 (2)
Biopsy performed, n (%)23/129 (18)
RCC13/23 (57)
Benign10/23 (43)
Postoperative day 1 eGFR (mL/min/1.73 m2), mean (sd)68.9 (27.8)
Creatinine follow-up (months), mean (sd)19.7 (18.2)
eGFR at most recent follow-up (mL/min/1.73 m2), mean (sd)65.1 (25.5)
Change in eGFR (mL/min/1.73 m2), mean (sd), 95% CI−5.2 (15.6), −7.9 to −2.5 (P = 0.03)
Patients with renal disease progression, n (%)*25/129 (20)

Of the 23 tumours biopsied, 57% showed malignant features. At a mean (sd) creatinine follow-up of 19.7 (18.2) months, mean (sd) postoperative eGFR was 65.1 (25.5) mL/min/1.73 m2, representing a mean (sd; 95% CI) decline in eGFR of 5.2 (15.6; −7.9 to–2.5) mL/min/1.73 m2 (P = 0.03). NKF-CKD stage progression occurred in 20% (25/129) patients. Multiple logistic regression analysis assessing variables predicting renal disease progression is summarized in Table 4. Significant predictors of progression included age ≥ 70 years (odds ratio [OR], 4.31, P = 0.03), hilar tumour location (OR, 4.67, P = 0.04) and postoperative day 1 eGFR ≤ 60 mL/min/1.73 m2 (OR, 7.09, P = 0.02).

Table 4. Multiple logistic regression analysis of renal disease progression.
VariableOdds ratio95% CIP
  1. CCI, Charlson comorbidity index; eGFR, estimated glomerular filtration rate.
Age ≥ 70 years4.311.15–10.20.03
Age-adjusted CCI ≥ 61.270.41–3.920.7
Nephrometry score ≥ 71.310.39–4.390.7
Tumour size ≥ 3 cm1.580.47–5.260.5
Endophytic growth2.880.66–12.40.2
Distance to collecting system ≤ 7 mm1.280.32–5.050.7
Hilar location4.671.05–20.80.04
Postoperative day 1 eGFR ≤ 60 mL/min/1.73 m27.091.40–35.70.02

The treatment success rate was 87% (112/129) at a mean (sd) follow-up of 30.2 (18.8) months. Kaplan–Meier curves for disease-free survival (DFS) and overall survival (OS) are shown in Fig. 1. No cancer-specific deaths occurred in our cohort, resulting in a cancer-specific survival of 100%. The Kaplan–Meier-estimated OS and DFS, respectively, were 95% and 89% at 1 year, 92% and 85% at 2 years, and 85% and 85% at 3 years. The characteristics of tumour recurrences are shown in Table 5. Tumour recurrence occurred in 17/124 (14%) patients at a mean (sd) of 8.7 (11.7) months after PCA. Mean (sd) tumour size and nephrometry score among treatment failures were 3.4 (1.2) cm and 7.3 (1.7), respectively. Of the treatment failures, 71% (12/17) occurred within 12 months and 94% (16/17) occurred within 16 months after PCA. Salvage therapy included repeat PCA (n = 10) and radical nephrectomy (n = 1); the remaining six patients chose imaging surveillance alone. Of the 10 patients undergoing repeat PCA, two secondary treatment failures occurred at 6.6 and 12.8 months. Cox proportional-hazards regression analysis evaluating predictors of disease recurrence is summarized in Table 6. Tumour size ≥ 3.0 cm was the only significant predictor of PCA failure (hazard ratio, 3.21, P = 0.03).

Figure 1.

Kaplan–Meier analyses for (A) disease-free survival and (B) overall survival. Of note, there were no cancer-specific deaths in the patient cohort.

Table 5. Summary of treatment failures.
Age (years)AA-CCITime to recurrence (months)Tumour size (cm)Nephrometry scoreRetreatmentOutcome
  1. AA-CCI, age-adjusted Charlson comorbidity index; NED, no evidence of disease; PCA, percutaneous cryoablation.
5680.033.37pPCANED to date
7150.834.510aPCANED to date
6751.103.49phPCANED to date
7151.435.05pPCANED to date
6291.606.17xNoneContinued surveillance
86101.805.39phPCANED to date
6741.903.47aPCANED to date
7052.002.15aPCAFailed, 12.8 months; underwent third PCA
6063.104.55pPCANED to date
85104.972.58pNoneContinued surveillance
62410.42.49pPCAFailed, 6.6 months; Continued surveillance
62511.03.35pRadical nephrectomyPathology: clear cell RCC NED to date
72714.22.910aNoneContinued surveillance
85814.43.36pNoneContinued surveillance
84915.41.97pPCANED to date
65915.52.67aNoneContinued surveillance
81548.21.98pNoneContinued surveillance
Table 6. Cox proportional-hazards regression analysis of tumour recurrence.
VariableHazard ratio95% CIP
Body mass index ≥ 30 kg/m22.600.89–7.600.08
Nephrometry score ≥ 71.680.50–5.630.4
Tumour size ≥ 3 cm3.211.09–9.440.03
Endophytic growth1.190.24–5.890.8
Distance to collecting system ≤ 7 mm1.060.30–3.820.9
Anterior location1.020.34–3.070.9
Hilar location1.610.43–6.080.5

Discussion

As the utilization of diagnostic abdominal imaging has continued to rise, the incidence of renal masses in elderly comorbid patients has also increased. Most of these elderly patients diagnosed with SRMs die from causes other than RCC [20]. Accordingly, the AUA guidelines recommend ablative procedures or active surveillance in those patients who are unfit for extirpative surgery [21]. Both LCA and PCA represent minimally-invasive cryoablative options. The oncological success for LCA has been shown in the short-, intermediate- and long-term, whereas intermediate and long-term studies after PCA are limited [9-13, 22-25]. We present what we consider to be the largest reported experience with PCA (129 tumours in 124 patients) with mean oncological and renal function follow-ups of 30 and 20 months, respectively.

Patient cohorts in both studies were similar with respect to mean age (73 and 72 years) and the proportion of patients undergoing PCA as a result of medical comorbidity (60% and 53%). The mean length of postoperative stay was similar in both studies (1.0 vs 1.3 days). Complication rates (6% and 9%) and blood transfusion requirements (3% vs 2%) were also similar.

In the present study, treatment success was achieved in 87% (112/129) of tumours at a mean follow-up of 30 months. Although our success rate falls within the range of other published reports, two studies reported higher success rates: 95% (88/93) at a mean follow-up of 26 months and 96% (52/54) at a mean follow-up of 21 months, respectively [26, 27]. Differences in PCA technique among these studies in comparison with the present study are probably key contributors to the disparity in treatment success. Atwell et al. [26] report the use of variable freeze times based on ice ball growth, whereas we performed freeze cycles at fixed 10-min intervals. Variable freeze time with the goal of ice ball extension 5 mm beyond the tumour margin may potentially improve the likelihood for PCA success, especially for larger tumours. Similarly, Vricella et al. [27] report the use of additional freeze–thaw cycles (up to four cycles) to ensure tumour destruction based on real-time CT images. Considering that our combined primary and secondary PCA success rates of 93% (120/129) approaches the success rates of Atwell et al. [26] and Vricella et al. [27], our current PCA technique of two cycles with fixed 10-min freeze times may be too conservative. Additionally, both Atwell et al. [26] and Vricella et al. [27] report the use of general anaesthesia in all patients. This may allow for improved tumour localization, resulting in fewer cryoprobe placement attempts and superior ice ball visualization.

The lower success rate in the present study compared to the study reported by Vricella et al. [27] may be a result of the difference in the number of tumours treated that were >3.0 cm. Our multivariable regression analysis showed that a tumour size > 3.0 cm was significantly associated with PCA treatment failure. Although mean tumour sizes were similar between the present study and the study reported by Vricella et al. [27] (2.7 vs 2.5 cm, respectively), only 19% (10/54) of tumours treated by Vricella et al. [27] were >3.0 cm, whereas 34% (44/129) tumours in the present study were >3.0 cm, which may have lead to an increased propensity for PCA failure.

Among studies of PCA with a follow-up >30 months, Malcolm et al. [13] and Miki et al. [28] report lower success rates than these encountered in the present study: 75% (15/20) at a mean follow-up of 30 months and 85% (11/13) at a mean follow-up of 35 months, respectively. Notably, Malcolm et al. [13] utilized a PCA technique similar to that used in the present study, potentially resulting in inadequate ice ball extension, as discussed above. The combined primary and secondary PCA success rate for Malcolm et al. [13] was 95% (19/20), which may suggest the need for a more aggressive PCA technique. The 85% treatment success rate at a mean follow-up of 35 months reported by Miki et al. [28] may indicate that other studies lack sufficient follow-up to capture late failures. However, 80% (4/5) of treatment failures reported by Atwell et al. [26] occurred within 3 months and 100% (5/5) occurred within 14 months. Additionally, 94% (16/17) of treatment failures in the present study occurred within 16 months, suggesting that most failures occur during relatively short follow-up.

Multivariable Cox proportional-hazards regression analysis in the present study showed that tumour size ≥ 3.0 cm was significantly associated with treatment failure (P = 0.03). A significant association between tumour recurrence and tumour size ≥ 3.0 cm has been reported in multiple studies of LCA [25, 29, 30]. However, Vricella et al. [27] found no variables that were significantly associated with PCA treatment failure on univariate analysis, including tumour size [27]. Intuitively, tumour size constraints for effective cryoablation should not be dependent on cryoablation modality. The study by Vricella et al. [27] may have been underpowered to detect significant differences between success and failure groups, considering a limited total sample size (n = 54), as well as a limited number of tumours ≥3.0 cm (n = 10). Additionally, modification of the PCA technique, with more liberal freeze times and freeze–thaw cycles, may increase the likelihood of treatment success for larger tumours.

Patients experienced a mean (sd) eGFR decline of 5.2 (15.6) mL/min/1.73 m2 at a mean follow-up of 20 months in the present study, which represents a minor yet statistically significant decline from preoperative eGFR (P = 0.03). Notably, the cohort of patients in the present study had significant baseline comorbidity (mean age-adjusted CCI of 6.4) and was probably predisposed to eGFR decline regardless of PCA intervention. Thus, comparing the eGFR decline in the present study with eGFR decline after other RCC treatment modalities would provide little clinically meaningful information as a result of significant differences in baseline patient characteristics. More appropriately, we performed a multivariable analysis, including both patient and procedural variables, aiming to identify factors associated with a decline in NKF-CKD disease stage [18]. Age ≥ 70 years (P = 0.03), hilar tumour location (P = 0.04) and postoperative day 1 eGFR ≤ 60 mL/min/1.73 m2 (P = 0.02) were significant predictors of kidney disease progression. Advanced age would probably predispose patients to renal disease progression regardless of PCA treatment. Although our experience of treating hilar tumours with PCA is small (only 11% of tumours were hilar in location), the significant association between hilar location and disease progression may be related to hilar and larger intraparenchymal vessels being damaged during PCA. Furthermore, the significant association of disease progression with postoperative day 1 eGFR suggests that damage to renal function sustained during PCA is a durable insult. Considering the intra-operative risks associated with PCA of hilar tumours and the possible association with long-term renal disease progression, patients with such tumours should be counselled on all treatment options, including surveillance.

The impact of PCA on the decline of renal function is most critical in patients with solitary kidney. In the present study, 11% (14/124) of patients undergoing PCA had undergone previous contralateral radical nephrectomy. This subset of patients experienced a statistically insignificant mean (sd) eGFR decline of 3.6 (17.4) mL/min/1.73 m2 (P = 0.45) from a mean (sd) baseline preoperative eGFR of 56.6 (21.0) mL/min/1.73 m2 at a mean follow-up of 23 months. Altunrende et al. [31] report similar results for 29 patients undergoing PCA for tumour in a solitary kidney. Their patients also experienced a statistically insignificant mean eGFR decline of 3.1 mL/min/1.73 m2 (P = 0.45) at a mean follow-up of 14 months. No cancer-specific deaths occurred in either the solitary kidney subset in the present study or in the patients treated by PCA in the study by Altunrende et al. [31]. This oncological outcome combined with limited renal function insult suggests that PCA is an appropriate alternative for patients presenting with renal tumours in a solitary kidney.

The limitations of the present study include its retrospective study design, relatively small sample size, limited follow-up, and a lack of pathological data in most cases. Although the present study represents the largest PCA series, the sample size limits interpretation of the analyses because of the low rate of treatment failure for PCA [13, 25-27]. Similarly, the present study includes one of the longest mean follow-up durations, although the slow growth of most renal tumours requires longer-term studies with even more substantial follow-up. The major limitation of the present study is the relatively few biopsies performed (18% of tumours). Similar to other studies of percutaneous ablative treatments, tumour control was based on radiographic studies without pathological confirmation. Considering that biopsies are not used to guide PCA treatment decisions, we assumed that they posed a risk of bleeding, leading to possible treatment complications and lower treatment success rates. Nevertheless, as our experience with PCA has grown, we now perform renal biopsies routinely after the placement of cryoprobes. Future studies should incorporate pathological data to facilitate the interpretation of oncological outcomes.

In a summary of our experience, PCA showed low intra-operative, postoperative and major complication rates, at the same time as achieving adequate oncological control and preservation of renal function. At a mean follow-up of 30 months, the PCA treatment success rate was 87%, and Kaplan–Meier-estimated cancer-specific survival, OS and DFS were 100%, 85% and 85% at 3 years. Tumour size ≥ 3.0 cm was the only significant predictor of treatment failure. Age ≥ 70 years, hilar tumour location and eGFR ≤ 60 mL/min/1.73 m2 on postoperative day 1 were significant predictors of kidney disease progression. PCA of renal masses represents a safe and oncologically acceptable alternative to surgery in patients with significant medical comorbidities. Baseline patient and tumour characteristics probably influence the risk of renal disease progression in patients undergoing PCA.

Source of Funding

Funded in part by a grant from Midwest Stone Institute.

Conflict of Interest

None declared.

Abbreviations
AUA

American Urological Association

CCI

Charlson comorbidity index

DFS

disease-free survival

eGFR

estimated glomerular filtration rate

LCA

laparoscopic cryoablation

NKF-CKD

National Kidney Foundation-Chronic Kidney Disease

OR

odds ratio

OS

overall survival

PCA

percutaneous cryoablation

SRMs

small renal masses

Ancillary