PATIENTS AND METHODS
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
Data were retrieved from seven European institutional databases of patients treated with either PN or RN for histologically confirmed RCC, between 1987 and 2007. Only patients with stages T1a–bN0M0 were selected for the present study (451 patients).
Clinical and pathological data, including preoperative American Society of Anesthesiology (ASA) score, GFR and symptom classification (asymptomatic, local or systemic symptoms) were gathered prospectively at each institution [5–12]. Preoperative GFR measurements were available for all patients and were derived from the Modification of Diet in Renal Disease study group equation . Clinical stage was assigned according to the 2002 clinical TNM staging system. The T stage and tumour size were based on CT. Absence of adenopathy was also confirmed radiographically. Finally, CT of the chest or chest X-rays complemented the evaluation. Other imaging procedures were used at the discretion of the treating physicians. The rate of PN vs RN was determined according to the operative reports. Histological subtypes were defined according to the 2002 Union Internationale Contre le Cancer classifications .
Independent-sample t-tests and chi-square tests were, respectively, used for comparisons of means and proportions. Univariate and multivariate competing-risks regression analyses tested the effect of the ASA score, GFR, T stage (T1a vs T1b) and nephrectomy type on RCC-specific mortality, as well as on non-RCC-related mortality as described by Fine and Gray .
Cumulative incidence plots were used to graphically depict the effects of ASA score, GFR, T stage and nephrectomy type on RCC-specific mortality and on non-RCC-related mortality. The date of nephrectomy was considered as the start of follow-up. All tests were two-sided with the significance level set at 0.05.
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
The characteristics of the 451 patients with T1a–bN0M0 RCC are shown in Table 1. The characteristics of patients with an ASA score of ≤2 and of patients with an ASA score of ≥3 were similar for gender, T stage and histologicl subtypes (all P > 0.05). Conversely, relative to patients with an ASA score of ≤2, patients with ASA scores of ≥3 were older (P < 0.001), had lower GFRs (P = 0.002), had smaller tumours (P = 0.01) and had lower Fuhrman grades (P < 0.001).
Table 1. Clinical characteristics of the study population of 451 patients stratified according to ASA score (= 2 vs. = 3)
|Variable||ASA score, n (%)||P|
|ASA score ≤2||ASA score ≥3||Overall|
|Total no. patients||348||103||451|| |
|Mean age, years|| 59.3|| 64.4|| 60.5||<0.001|
|Gender:|| || || ||0.09|
| Male||226 (64.9)|| 76 (73.8)||302 (67.0)|| |
| Female||122 (35.1)|| 27 (26.2)||149 (33.0)|| |
|Symptom classification:|| || || ||<0.001|
| Asymptomatic||246 (70.7)|| 55 (53.4)||301 (66.7)|| |
| Local symptoms|| 64 (18.4)|| 16 (15.5)|| 80 (17.7)|| |
| General symptoms|| 5 (1.4)|| 32 (31.1)|| 37 (8.2)|| |
| Unknown|| 33 (9.5)|| 0 (0.0)|| 33 (7.3)|| |
|GFR, mL/min:|| || || ||0.002|
| >60||257 (73.9)|| 59 (57.3)||316 (70.1)|| |
| ≤60|| 91 (26.1)|| 44 (42.7)||135 (29.9)|| |
|Mean tumour size, cm|| 3.8|| 3.2|| ||0.01|
|T stage:|| || || ||0.1|
| T1a||242 (69.5)|| 80 (71.2)||322 (71.4)|| |
| T1b||106 (30.5)|| 20 (22.3)||129 (28.6)|| |
|Histological subtype:|| || || ||0.2|
| Clear cell||227 (65.2)|| 71 (68.9)||298 (66.1)|| |
| Papillary|| 71 (20.4)|| 25 (24.3)|| 96 (21.3)|| |
| Other|| 36 (4.6)|| 5 (4.9)|| 41 (9.1)|| |
| Unknown|| 14 (4.0)|| 2 (1.9)|| 16 (3.5)|| |
|Fuhrman grade:|| || || ||<0.001|
| Low (1 or 2)||224 (64.4)|| 88 (85.4)||312 (69.2)|| |
| High (3 or 4)||124 (35.6)|| 15 (14.6)||139 (30.8)|| |
|Nephrectomy type:|| || || ||<0.001|
| PN||240 (69.0)|| 99 (96.1)||339 (75.2)|| |
| RN||108 (31.0)|| 4 (3.9)|| 112 (24.8)|| |
|Mean (range) follow-up, years|| 3.1 (0.1–17.8)|| 3.2 (0.1–13.9)|| ||0.001|
Figure 1A–D represent cumulative incidence plots which depict RCC-specific, as well as non-RCC-related mortality for patients with T1a–b RCC treated with either PN or RN. During follow-up, eight of 451 patients died of RCC vs 24 of 451 who succumbed to non-RCC-related causes. At 2 and 5 years, the follow-up was of 643 and 1156 patient years.
Figure 1. Cumulative incidence plots depicting the RCC-specific mortality and other cause in the cohort (451 patients). Data are stratified according to ASA score (≤2 vs ≥3, A), GFR (≤60 vs > 60 mL/min per 1.73 m2, B), T stage (T1a vs T1b, C) and nephrectomy type (PN vs RN, D).
Download figure to PowerPoint
Mortality according to preoperative ASA score
The competing-risks regression-derived RCC-specific mortality rates in patients with T1a-b RCC and with an ASA score of ≤2 after adjustment for non-RCC-related mortality were 0.4, 0.8 and 0.8% at, respectively, 1, 2 and 5 years. Conversely, for the same time points non-RCC-related mortality rates were 0.7, 0.7 and 4.9%. The competing-risks regression-derived RCC-specific mortality rates in patients with T1a–b RCC and with ASA scores of ≥3 after adjustment for non-RCC-related mortality were 2.0, 4.9 and 4.9% at, respectively, 1, 2 and 5 years. Conversely, for the same time points non-RCC-related mortality rates were 4.2, 4.2 and 14.4%.
Mortality according to baseline GFR
For patients with T1a–b RCC and a GFR of >60 mL/min treated either with PN or RN, the RCC-specific mortality rates at, respectively, 1, 2 and 5 years were 0.7, 2.2 and 2.2% vs 0.8, 0.8 and 3.7% for non-RCC-related mortality. For patients with T1a–b RCC and a GFR of ≤60 mL/min, the RCC-specific mortality rates at, respectively, 1, 2 and 5 years were 0.9, 0.9 and 0.9% vs 3.5, 3.5 and 15.2% for non-RCC-related mortality.
Mortality according to T stage
For patients with T1a RCC treated either with PN or RN, the RCC-specific mortality rates at, respectively, 1, 2 and 5 years were 0.7, 2.2 and 2.2% vs 1.2, 1.2 and 6.5% for non- RCC-related mortality. For patients with T1b RCC, the RCC-specific mortality rates at the same time points were 0.9, 0.9 and 0.9% vs 2.6, 2.6 and 9.4% for non-RCC-related mortality.
Mortality according to treatment type
For patients with T1a–b RCC treated with PN, the RCC-specific mortality rates at, respectively, 1, 2 and 5 years were 0.7, 2.1 and 2.1% vs 1.4, 1.4 and 7.3% for non-RCC-related mortality. For patients with T1a–b RCC treated with RN, the RCC-specific mortality rates at, respectively, 1, 2 and 5 years were 1.0, 1.0 and 1.0% vs 2.3, 2.3 and 7.2% for non-RCC-related mortality.
Table 2 shows the multivariate competing-risks regression models. In the model addressing RCC-specific mortality after accounting for non-RCC-related mortality, only the ASA score reached independent predictor status. In the model addressing the non-RCC-related mortality after accounting for RCC-specific mortality, the ASA score and GFR reached independent predictor status in multivariate analyses.
Table 2. Multivariate competing-risks regression models addressing RCC-specific mortality (after accounting for non-RCC-related mortality in the entire cohort of 451 patients) and non-RCC-related mortality (after accounting for RCC-specific mortality in the entire cohort)
| ||Multivariate P|
|ASA score, ≥3 vs ≤2||0.003|
|GFR (mL/min per 1.73 m2), ≤60 vs >60||0.7|
|Nephrectomy type: RN vs PN||0.9|
|T stage, T1b vs T1a||0.2|
|ASA score, ≥3 vs ≤2||0.004|
|GFR (mL/min per 1.73 m2), ≤60 vs >60||0.004|
|Nephrectomy type: RN vs PN||0.4|
|T stage, T1b vs T1a||0.9|
Table 3 shows the characteristics of the eight RCC-specific deaths, who died from RCC according to T1 substages. It is noteworthy that five of the eight patients had Fuhrman grades 1 and 2 RCC.
Table 3. Characteristics of individual patients who died from RCC according to T1 substages
|Patients||Age, years||Gender||Symptom classification||ASA score||GFR, mL/min||Tumour size, cm||Treatment type||Histological subtypes||Fuhrman grade|
|Patient 1||73||Male||Asymptomatic||4||53.1||3.0||PN||Clear cell||3|
|Patient 4||67||Male||Asymptomatic||3||68.7||2.8||PN||Clear cell||1|
|Patient 5||62||Male||Asymptomatic||3||90.3||2.0||PN||Clear cell||2|
|Patient 1||77||Male||Asymptomatic||3||41.9||5.5||PN||Clear cell||1|
|Patient 2||77||Male||Local symptoms||3||45.0||6.5||PN||Papillary||2|
|Patient 3||59||Female||Local symptoms||1||81.5||4.4||RN||Clear cell||4|
- Top of page
- PATIENTS AND METHODS
- CONFLICT OF INTEREST
European data on RCC-specific mortality adjusted for non-RCC-related mortality are scarce. Competing-risks regression models provide the ideal modelling technique for the most bias-free assessment of RCC-specific mortality. Competing-risks regression models have only been used in one analysis of North America data . Based on the paucity of European analyses that rely on competing-risks regression models we applied this methodology to a large multi-institutional database. Our objective was to determine the RCC-specific mortality rates for T1a and T1b tumours. Moreover our aim was to provide European data that could be used to support the existing follow-up schemes of the EAU guidelines.
We used competing-risks regressions, as described by Fine and Gray, as this modelling technique is capable of adjustment for non-cancer-related mortality . Competing causes of mortality represent an important consideration in T1a–b RCC, due to the relatively indolent course of most such tumours. The use of Cox regression models in circumstances when non-cancer-related mortality claims a large proportion of death usually leads to an important overestimation of cause-specific mortality . This in turn results in biased estimates, away from the null. The bias related to non-RCC-related mortality may be effectively controlled for, if competing-risks regression modelling is used. Based on this important methodological consideration, the present results were entirely based on competing-risks regression models. The present findings show that five of 322 T1a and three of 129 T1b patients died of RCC. This resulted in RCC-specific mortality rates of 0.7, 2.2 and 2.2% at, respectively, 1, 2 and 5 years for patients with T1a RCC and in RCC-specific mortality rate of 0.9, 0.9 and 0.9% at the same time points for patients with T1b tumours.
Interestingly, of patients with T1a–b five of the eight RCC-specific deaths were caused by low Fuhrman grade (1 or 2) RCC. Although the very few events limit the generalizability of the present results, the present data indicate that patients with T1a–b RCC and with low Fuhrman grade may die of RCC. In consequence, low Fuhrman grade may not represent a valid indicator for surveillance in patients with small renal masses [10,12,17,18].
The virtually same rates of RCC-specific mortality for T1a and T1b patients also indicate that T1 substages cannot be regarded as a reliable predictor of RCC mortality. Incidentally, the variable defining T1 substages did not achieve independent predictor status in the multivariate competing-risks analyses.
Finally and most importantly, the ASA score was an independent predictor of cancer-specific and non-cancer-related mortality. Expectedly, non-RCC-related mortality claimed three-times more lives than cancer-specific mortality in patients with ASA scores of ≥3. The non-RCC-related mortality rates were 0.7, 0.7 and 4.9% vs 4.2, 4.2 and 14.4% at 1, 2 and 5 years respectively, for ASA scores of ≤2 vs ≥3 in T1a–bN0M0 RCC. The non-RCC-related mortality data are equally important to RCC-specific mortality data and show the strength of the effect of the ASA score on non-RCC-related mortality. They indicate that surgical management represents a valid option for ASA scores of 1 or 2 with T1a–b RCC. Conversely, for patients with ASA scores of ≥3, the relatively elevated non-RCC-related mortality rates indicate that alternative treatments such as cryotherapy, radiofrequency or other non-extirpative local tumour destruction techniques may be more valid. This finding is consistent with the indications for surveillance. This option is suggested for patients with T1a lesions and important comorbidities . The present data suggest that this indication may be expanded to T1b patients with ASA scores of ≥3 as their risk of RCC-specific mortality was 0.9, 0.9 and 0.9% vs 2.6, 2.6 and 9.4% for non-RCC-related mortality.
North American data corroborate the present findings . For example, Hollingsworth et al. also reported that mortality after nephrectomy was negligible for small renal masses. Conversely, they also reported an elevated rate of non-cancer-related mortality, especially in patients with baseline comorbidities.
We were unable to replicate the findings of Thompson et al. regarding the effect of PN vs RN on non-cancer-related mortality. More limited follow-up of our series (3 years) relative to the Thompson et al. series (7 years) represent a cause. However, it also possible that the rate of PN use relative to RN (higher in the present series 75.2 vs 57.2%) differed between the two cohorts and accounted for the recorded lack of difference.
The limitation of our finding consisted of the relative few events. However, at 1, 2 and 5 years, 368, 643 and 1156 person year of follow-up were available. Consequently, it can be postulated that the low event rate is not due to sample or follow-up limitations, but relates to the indolent natural history of treated T1a–b RCC.
In conclusion, the present data indicate that a GFR of ≤60 mL/min per 1.73 m2 and/or an ASA score of ≥3 affect mortality and that both variables need to be interpreted as relative contra-indications to definitive management of T1a–b RCC. Moreover, the present findings show that nephrectomy virtually eliminates the cancer-specific mortality risk in patients with T1a–b RCC. The present data validate the EAU follow-up recommendations for this patient category.