- Email: [email protected]
Determination of the Cutoff Value of the Proportion of Cystic Change for Prognostic Stratification of Clear Cell Renal Cell Carcinoma
Determination of the Cutoff Value of the Proportion of Cystic Change for Prognostic Stratification of Clear Cell Renal Cell Carcinoma Heae Surng Park, Kyoungbun Lee and Kyung Chul Moon* From the Department of Pathology (HSP, KL) and Kidney Research Institute, Medical Research Center, Seoul National University College of Medicine (KCM), Seoul, Republic of Korea
Purpose: Cystic renal cell carcinoma has more favorable biology than noncystic renal cell carcinoma. Recently cystic change detected grossly or by low power microscopy was found to be a good prognostic factor for clear cell renal cell carcinoma. We assessed the optimal cutoff value of the proportion of cystic change with prognostic significance for clear cell renal cell carcinoma. Materials and Methods: We identified 223 patients with clear cell renal cell carcinoma who underwent partial or radical nephrectomy between 2001 and 2003. The cystic proportion of the tumor cut surface was calculated objectively and its prognostic significance was evaluated. Results: The ROC curve showed that a cystic percent of between 6% and 10% was appropriate to detect patients with renal cell carcinoma at low risk for cancer mortality and progression. A cutoff of 6% was adopted as a break point of cystic change for patient stratification. We analyzed the records of 87 patients (39.0%) with tumors with a cystic proportion of greater than 5%, that is 6% or greater. They had significantly lower stage and lower Fuhrman nuclear grade than patients with tumors with a cystic change of 5% or less (each p ⬍0.0001). On multivariate analysis a cystic proportion of more than 5% was a good prognostic indicator of cancer specific and progression-free survival (HR 0.221, p ⫽ 0.044 and HR 0.214, p ⫽ 0.004, respectively). Conclusions: In patients with clear cell renal cell carcinoma a cystic change comprising more than 5% of the tumor is a good independent predictor of survival.
Abbreviations and Acronyms ccRCC ⫽ clear cell RCC MCRCC ⫽ multilocular cystic RCC RCC ⫽ renal cell carcinoma Submitted for publication October 28, 2010. Study received Seoul National University Hospital institutional review board approval. Supported by the Basic Science Research Program through the National Research Foundation of Korea, funded by Ministry of Education, Science and Technology Grant 2010-0004550. * Correspondence: Department of Pathology and Kidney Research Institute, Medical Research Center, Seoul National University College of Medicine, 28 Chongno-gu Yeongeon-dong, Seoul 110-799, Republic of Korea (telephone: 82-2-20721767; FAX: 82-2-7432-5530; e-mail: blue7270@ snu.ac.kr).
Key Words: kidney; carcinoma, renal cell; cysts; mortality; Republic of Korea CYSTIC change is not rare in RCC.1 This feature was found in 4% to 15% of RCCs in an imaging study2 and in approximately 30% of ccRCCs examined in a recent histopathological review.3 RCC with a predominantly cystic pattern of growth, defined as neoplastic cells occupying 25% or less of tumor volume, was considered cystic RCC. Some groups have reported that cystic RCC has more favorable biology than noncystic RCC.3– 8 Currently
MCRCC, which is composed of an entirely cystic lesion, is recognized as a distinct subtype of ccRCC with excellent prognosis in the 2004 WHO classification of tumors.9 –11 Previously we reported that a cystic change detected on gross examination or low power microscopy was a good prognostic factor for ccRCC despite the exclusion of MCRCC.3 To our knowledge no group has systematically examined the association
0022-5347/11/1862-0423/0 THE JOURNAL OF UROLOGY® © 2011 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION
Vol. 186, 423-429, August 2011 Printed in U.S.A. DOI:10.1016/j.juro.2011.03.107
AND
RESEARCH, INC.
www.jurology.com
423
424
RENAL CELL CANCER WITH SIGNIFICANT CYSTIC CHANGE
between the proportion of cystic change and prognosis in patients with ccRCC. In this study we determined an optimal cystic proportion cutoff value to define 2 groups of patients with ccRCC with statistically significantly different outcomes.
MATERIALS AND METHODS Patient Selection We reviewed the pathology slides of consecutively resected ccRCCs between 2001 and 2003 at Seoul National University Hospital. MCRCC or RCC with features suggesting clear cell papillary RCC were excluded from study. MCRCC was defined according to 2004 WHO criteria as a tumor entirely composed of multilocular cysts lined by thin septa containing clear tumor cells with no expansile solid nodule on gross or microscopic examination.12 Clear cell papillary RCC is a distinctive group of renal tumors composed mainly of cells with clear cytoplasm arranged in cystic and papillary patterns. Clear cell papillary RCC is confined to the kidney with no recurrence or metastasis.13,14 Also excluded from study were ccRCC in patients with von Hippel-Lindau syndrome, dialysis associated cystic RCC and metachronous ccRCC. A total of 223 patients with ccRCC were identified for analysis. This study was approved by the Seoul National University Hospital institutional review board. All slides were reviewed by 2 pathologists.
Clinicopathological Evaluation The clinical features included patient age at surgery, gender, surgery type and clinical outcome. Followup was defined as the interval between the date of nephrectomy and the date of death or last followup. The pathological features evaluated were tumor size, TNM stage according to the 2009 American Joint Committee on Cancer,15 regional
lymph node involvement, distant metastasis, nuclear grade according to the Fuhrman system and predominant cyst fluid content. If no pathological confirmation of metastatic disease was performed, patients were assessed for clinical metastatic stage based on clinical examination and radiological studies. Tumor size was measured as the longest dimension. Nuclear grade was based on the highest grade tumor area identified. Cysts were considered dilated spaces filled with amorphous fluid or red blood cells lined by clear tumor cells with a dimension of at least 0.1 cm (fig. 1, A to C). However, cases in which the tumor contained cystic spaces containing necrotic tumor tissue were excluded from study since RCC with cystic necrosis carries a worse prognosis (fig. 1, E).16 A lumen filled with fluid in a hyalinized area was also considered a cyst. The predominant fluid content in the cysts was categorized as predominantly serous with amorphous eosinophilic fluid accounting for 60% or greater of the whole cystic content, predominantly bloody with red blood cells accounting for 60% or greater of the whole cystic content and mixed. Red blood cell extravasation between tumor cells was not considered a cyst filled with blood (fig. 1, D).
Cystic Proportion Determination Hematoxylin and eosin glass slides of tumors of each case were scanned using a Super Coolscan® 9000 ED film scanner. The digitalized slide images and digital gross pictures were used to quantify the area of the lumina in the cystic component and the total area of the tumor cut surface with ImageJ (fig. 2). The proportion of the cystic component was calculated by dividing the sum of cystic luminal areas by the whole area of the tumor cut surface. Tumors with a cystic component of less than 1% were considered noncystic RCC (0% cystic). There were 2 potential errors when determining the cystic proportion in this study. 1) Operator error could
Figure 1. RCC features. A, gross view of small RCC shows no apparent cysts. B, microscopy of same tumor reveals many microcysts containing predominantly serous fluid (scan power). C, RCC with predominantly bloody cysts. Reduced from ⫻40. D, RCC with extravasation of red blood cells. Reduced from ⫻100. E, RCC with cystic necrosis. Reduced from ⫻40.
RENAL CELL CANCER WITH SIGNIFICANT CYSTIC CHANGE
425
RESULTS
Figure 2. Measurement of cystic and total tumor areas. Luminal space inside each cyst (light blue outlines) and tumor were selected as regions of interests. Areas of regions were measured by pixel count. Cystic component of tumor in this section is 8%.
have increased the cystic proportion when more cystic areas of tumor were sectioned to generate the pathology slides. We believe that this error was minor since gross specimens and slides were subsequently reviewed. 2) There was potential measurement error. We found cystic proportion differences of 1% to 2% when the cystic portion of the same tumor image was measured by the same individual using ImageJ software. To overcome the measurement error we clustered RCC cases based on the percent of cystic area, including 0% (entirely solid), 1% to 5%, 6% to 10%, 11% to 15%, 16% to 20%, 21% to 25%, 26% to 30%, 31% to 35%, 36% to 40% and greater than 40%.
Statistical Analysis We calculated the area under the ROC curve according to the cystic proportion as a measure to predict a low risk of cancer specific death and progression. The optimal ROC curve threshold matched the cystic percent cutoff point showing improved sensitivity and specificity. An AUC of 0.5 to 0.7 indicated a low accuracy test, between 0.7 and 0.9 indicated moderate accuracy and more than 0.9 indicated high accuracy.17 Patients were classified into 2 groups by the identified optimal cutoff value. The Pearson chi-square and Student t tests were used to compare categorical and continuous variables, respectively. Cancer specific survival was measured from the date of surgery to the date of cancer related death or last followup. Progression-free survival was calculated from the date of surgery to the date of tumor recurrence/metastasis or last followup. Survival analysis was performed using the Kaplan-Meier method with results compared using the log rank test. Multivariate analysis was performed with Cox proportional hazards regression models. All statistical analysis was done using SPSS®, version 17.0 with p ⬍0.05 considered statistically significant.
Table 1 lists patient characteristics. Median age at surgery was 55 years (range 28 to 82) and 75.3% of the patients were male. Median followup was 64 months (range 2 to 97). Of the 223 patients 33 (14.8%) died of RCC and 12 (5.4%) died of another cause. A total of 164 patients (73.5%) had no evidence of RCC and 14 (6.3%) had disease progression. Median tumor size on pathological examination was 4 cm (range 1 to 16). In all tumors the size of the cyst was 0.1 to 4 cm. Cyst fluid contents were predominantly serous in 74 cases (50.0%), bloody in 49 (33.1%) and mixed in 25 (16.9%). The RCC cystic proportion was 0% to 80% (median ⫾ SD 9% ⫾ 14%). Of the tumors 88 (39.5%) were entirely solid, 48 (21.5%) had cysts taking up 1% to 5% of the tumor and 87 (39.0%) RCCs had a cystic proportion that accounted for more than 5% of the tumor. ROC curve analysis showed that the cystic component ratio had moderate accuracy to predict a low risk of cancer related death (mean AUC 0.742 ⫾ 0.042, 95% CI 0.660 – 0.823) and disease progression (AUC 0.745 ⫾ 0.037, 95% CI 0.673– 0.817). ROC curve data revealed that a cystic proportion of Table 1. Clinical and pathological characteristics of 223 patients with ccRCC Median ⫾ SD age (range) No. gender (%): F M Median ⫾ SD cm tumor size (range) No. nephrectomy (%): Partial Radical No. TNM stage (%): 1 2 3 4 No. Fuhrman nuclear grade (%): 1 2 3 4 No. predominant cystic content (%): Serous Bloody Mixed Median ⫾ SD % cystic component (range) No. cystic component (%): 0% 1%–5% 6%–10% 11%–15% 16%–20% 21%–25% 26%–30% 31%–35% 36%–40% Greater than 40%
55 ⫾ 11.2 (28–82) 55 (24.7) 168 (75.3) 4 ⫾ 3.1 (1–16) 32 191
(14.3) (85.7)
153 22 26 22
(68.6) (9.9) (11.6) (9.9)
22 102 76 23
(9.9) (45.7) (34.1) (10.3)
74 (50%) 49 (33.1%) 25 (16.9%) 9 ⫾ 14 (0–80) 88 48 23 15 9 10 7 10 3 10
(39.5) (21.5) (10.3) (6.7) (4.1) (4.5) (3.1) (4.5) (1.3) (4.5)
426
RENAL CELL CANCER WITH SIGNIFICANT CYSTIC CHANGE
Table 2. ROC curve data on sensitivity and specificity of each cystic cutoff No Ca Specific Death % Cystic Cutoff 1–5 6–10 11–15 16–20 21–25 26–30 31–35 36–40 41 or Greater
No Progression
% Sensitivity
% Specificity
Likelihood Ratio
% Sensitivity
% Specificity
Likelihood Ratio
66.8 44.7 32.6 24.7 20 15.3 11.6 6.8 5.3
75.8 93.9 93.9 93.9 93.9 97 97 100 100
2.76 7.33 5.34 4.05 3.28 5.1 3.87 . .
69.3 47.2 34.7 26.7 21.6 16.5 12.5 7.4 5.7
72.3 91.5 93.6 95.7 95.7 97.9 97.9 100 100
2.5 5.55 5.42 6.21 5.02 7.86 5.95 . .
between 6% and 10% was appropriate to identify patients with RCC at no risk for cancer specific death or disease progression (table 2). Sensitivity for no cancer related death and progression-free survival was 44.7% and 47.2% with 93.9% and 91.5% specificity, respectively. We adopted a cystic portion of 6% as the optimal cutoff to distinguish 2 groups of patients with RCC who had significantly different outcomes. A total of 136 patients (61.0%) had tumors with a cystic proportion of 5% or less while 87 (39.0%) had tumors with a cystic proportion of greater than 5% (table 3). RCCs with a cystic component of more than 5% of the tumor had lower stage and lower Fuhrman nuclear grade (each p ⬍0.0001). RCCs with cystic lesions of more than 5% of the tumor were associated with significantly longer cancer specific and progression-free survival than those with cystic lesions of 5% or less of the tumor (log rank p ⬍0.0001, fig. 3). On univariate analysis lower stage, lower Fuhrman nuclear grade, and a cystic proportion of more than 2% and 10% were also predictive variables of cancer related mortality and progression. However, the predominant cystic fluid content was not significantly related to patient outcome on survival analysis. Multivariate analysis revealed that a cystic proportion of greater than 5% (model 2) was an indeTable 3. Clinicopathological characteristics of patients with RCC by 5% cutoff % Cystic Proportion
No. pts (%) Median age No. gender (%): F M No. stage (%): I–II III–IV No. Fuhrman nuclear grade (%): 1–2 3–4
5 or Less
Greater Than 5
136 (61) 58
87 (39) 51
38 (27.9) 98 (72.1)
17 (19.5) 70 (80.5)
95 (69.9) 41 (30.1)
80 (92) 7 (8)
70 (51.5) 66 (48.5)
54 (62.1) 33 (37.9)
p Value 0.429 0.156
⬍0.0001 ⬍0.0001
pendent, good prognostic variable for cancer specific and progression-free survival (HR 0.221, 95% CI 0.051– 0.963, p ⫽ 0.044 and HR 0.214, 95% CI 0.054 – 0.613, p ⫽ 0.004, respectively), as were stage and Fuhrman nuclear grade (table 4). A cystic proportion of greater than 2% (model 1) and greater than 10% (model 3) were independent, good prognostic predictors of progression-free survival but they were not statistically significant for cancer specific survival on multivariate analysis. To determine whether cystic change contributes a survival benefit over tumor stage and nuclear grade we performed stratified survival analysis (fig. 4). A cystic proportion of more than 5% of tumor stratified by tumor stage was associated with better cancer specific and progression-free survival (p ⫽ 0.0140 and 0.0012, respectively, fig. 4, A and B). A cystic proportion of more than 5% of tumor stratified by nuclear grade also showed additional survival benefit for cancer specific and progression-free survival (p ⫽ 0.0001 and ⬍0.0001, respectively, fig. 4, C and D).
DISCUSSION Conventional RCC predominantly composed of multilocular cysts is considered cystic RCC. Cystic RCC reportedly accounts for 2.8% to 12% of RCCs and is associated with more favorable outcomes, according to some groups.4,7– 8 Since cystic RCC has been defined as a tumor with a cystic component representing at least 75% of the tumor,5–7 these studies may have included MCRCCs, which has a cystic component of almost 100%. MCRCC is currently defined by strict morphological criteria according to the 2004 WHO classification of tumors and no tumor with this feature has ever recurred or metastasized.12 MCRCC is differentiated from extensively cystic ccRCC by the absence of a grossly solid area or an expansile tumor cell mass on microscopy.18 In some studies the term cystic RCC includes MCRCC.4,8 Thus, the frequency of good outcomes of RCC with cystic change may possibly have been overestimated.
RENAL CELL CANCER WITH SIGNIFICANT CYSTIC CHANGE
427
Figure 3. Kaplan-Meier survival curves by cystic component with 5% cutoff for all patients (log rank test for trend p ⬍0.0001). A, cancer specific survival. B, progression-free survival.
We recently reported that a cystic change observed grossly or by low power microscopy was still statistically significantly associated with a lower risk of cancer progression in patients with ccRCC even when excluding those with MCRCC.3 Since MCRCC carries a good prognosis and RCC with cystic change also has a better prognosis than solid RCC, it may be hypothesized that patient survival is proportional to the percent of the cystic component and a cutoff point separates groups with significantly different prognoses. To our knowledge we report the first study of the association between the proportion of cystic change in ccRCC and clinical outcome. In previous studies the cystic component was calculated by subtracting the solid component from
total tumor area.5–7 This method does not reflect real tumor cell-free volume since neoplastic cells exist around the cysts. In this study to specifically select the cancer-free area, we defined the cystic component as the area inside the cyst. Since assessment of the cystic proportion by microscopy alone was not reproducible, pixel counts of the area of the cyst lumina and of the total tumor mass in the same digitalized image were measured using ImageJ software and the percent of the cystic component was calculated using those values. The ROC curve determined that cysts were a moderate predictor of a good clinical outcome in patients and a cystic proportion of between 6% and 10% was adequate to identify patients with greater
Table 4. Multivariate models by cystic proportion cutoff Proportion Hazards Model (prognostic variables) Model 1: Stage Nuclear grade % Cystic proportion Model 2: Stage Nuclear grade % Cystic proportion Model 3: Stage Nuclear grade % Cystic proportion Model 4: Stage Nuclear grade % Cystic proportion
Ca Specific Survival Category
Progression-Free Survival
HR (95% CI)
p Value
HR (95% CI)
p Value
cutoff
I–II vs III–IV 1–2 vs 3–4 2 or Less vs greater than 2
12.862 (4.914–33.667) 4.611 (1.540–13.805) 0.451 (0.177–1.149)
⬍0.0001 0.006 0.095
7.358 (3.743–14.467) 4.478 (2.007–9.989) 0.328 (0.154–0.700)
⬍0.0001 ⬍0.0001 0.004
cutoff
I–II vs III–IV 1–2 vs 3–4 5 or Less vs greater than 5
11.63 (4.467–30.284) 4.2 (1.390–12.691) 0.221 (0.051–0.963)
⬍0.0001 0.011 0.044
7.129 (3.637–13.976) 4.181 (1.865–9.371) 0.214 (0.074–0.613)
⬍0.0001 0.001 0.004
cutoff
I–II vs III–IV 1–2 vs 3–4 10 or Less vs greater than 10
14.059 (5.479–36.077) 4.072 (1.350–12.282) 0.359 (0.083–1.556)
⬍0.0001 0.013 0.171
8.353 (4.298–16.234) 3.879 (1.735–8.669) 0.263 (0.080–0.864)
⬍0.0001 0.001 0.028
cutoff
I–II vs III–IV 1–2 vs 3–4 20 or Less vs greater than 20
15.99 (6.341–40.317) 4.29 (1.433–12.846) 0.546 (0.128–2.324)
⬍0.0001 0.009 0.413
9.864 (5.034–18.628) 3.867 (1.743–8.579) 0.293 (0.070–1.221)
⬍0.0001 0.001 0.092
428
RENAL CELL CANCER WITH SIGNIFICANT CYSTIC CHANGE
Figure 4. Stratified Kaplan-Meier survival curves. A, cancer specific survival adjusted for stage (integrated over strata log rank test p ⫽ 0.0140). B, progression-free survival adjusted for stage (integrated over strata log rank test, p ⫽ 0.0012). C, cancer specific survival adjusted for nuclear grade (integrated over strata log rank test p ⫽ 0.0001). D, progression-free survival adjusted for nuclear grade (integrated over strata log rank test p ⬍0.0001).
than 90% specificity and with a high likelihood ratio. A cutoff of 6% was adopted as the break point of cystic change to stratify cases. RCC with a cystic proportion of greater than 5% was significantly associated with lower stage and lower nuclear grade than RCC with a cystic proportion of 5% or less. This result implies that tumors with a cystic change accounting for a relatively small portion of the whole tumor nodule also have favorable biology. Univariate analysis revealed that cystic change based on the 2%, 5% and 10% cutoffs was a statistically significant, good prognostic predictor of cancer specific and progression-free survival. However, only cystic change according to the 5% cutoff significantly decreased the risk of cancer specific death and disease progression on multivariate analysis. Significant survival differences between RCCs with a greater than 5% cystic lesion and RCCs with a cystic lesion of 5% or less were also seen in subgroups after stratifying patients into favorable (stages I–II and grades 1–2) and unfavorable (stages III–IV and grades 3– 4) groups. Thus, a cystic proportion of more than 5% is a favorable predictive factor regardless of stage or nuclear grade in ccRCC cases. The p values for correlations between the cystic percent and outcomes on univariate and multivariate analyses tended to increase as the cutoff increased. This tendency could have been the result
of grouping cases with a moderate amount of cystic area, such as 15%, but less than the cutoff, such as 20%, in the no cystic change group, although they had a good prognosis. Additional studies in large populations with much longer followup are required to clarify the relationship between cystic change and cancer prognosis. Cysts are occasionally seen during gross examination of the ccRCC specimen. Sometimes cysts are only apparent after microscopic examination due to obscuring cyst fluid or small cyst size. The presence of cysts is not enough to predict survival until they make up 5% of the tumor. Careful gross and microscopic examinations are needed to detect ccRCC with clinically significant cysts.
CONCLUSIONS Cystic change in ccRCC can be described by dilated cystic spaces with a diameter of greater than 0.1 cm regardless of cystic fluid content, except for necrotic tumor tissue. A cystic change of more than 5% of the tumor is an independent, good prognostic factor in patients with ccRCC. Careful pathological examination to detect ccRCC with clinically significant cystic change is needed. Future studies are needed to ascertain the pathogenesis of cystic change.
RENAL CELL CANCER WITH SIGNIFICANT CYSTIC CHANGE
429
REFERENCES 1. Grignon DJ and Che M: Clear cell renal cell carcinoma. Clin Lab Med 2005; 25: 305. 2. Hartman DS, Davis CJ Jr, Johns T et al: Cystic renal cell carcinoma. Urology 1986; 28: 145. 3. Park HS, Jung EJ, Myung JK et al: The prognostic implications of cystic change in clear cell renal cell carcinoma. Korean J Pathol 2010; 44: 149. 4. Bielsa O, Lloreta J and Gelabert-Mas A: Cystic renal cell carcinoma: pathological features, survival and implications for treatment. BJU 1998; 82: 16. 5. Corica FA, Iczkowski KA, Cheng L et al: Cystic renal cell carcinoma is cured by resection: a study of 24 cases with long-term followup. J Urol 1999; 161: 408. 6. Nassir A, Jollimore J, Gupta R et al: Multilocular cystic renal cell carcinoma: a series of 12 cases and review of the literature. Urology 2002; 60: 421. 7. Han KR, Janzen NK, McWhorter VC et al: Cystic renal cell carcinoma: biology and clinical behavior. Urol Oncol 2004; 22: 410.
8. Webster W, Thompson R, Cheville J et al: Surgical resection provides excellent outcomes for patients with cystic clear cell renal cell carcinoma. Urology 2007; 70: 900. 9. Imura J, Ichikawa K, Takeda J et al: Multilocular cystic renal cell carcinoma: a clinicopathological, immuno- and lectin histochemical study of nine cases. APMIS 2004; 112: 183. 10. Suzigan S, Lopez-Beltran A, Montironi R et al: Multilocular cystic renal cell carcinoma: a report of 45 cases of a kidney tumor of low malignant potential. Am J Clin Pathol 2006; 125: 217. 11. Gong K, Zhang N, He Z et al: Multilocular cystic renal cell carcinoma: an experience of clinical management for 31 cases. J Cancer Res Clin Oncol 2008; 134: 433. 12. Lopez-Beltran A, Scarpelli M, Montironi R et al: 2004 WHO classification of the renal tumors of the adults. Eur Urol 2006; 49: 798. 13. Gobbo S, Eble JN, Grignon DJ et al: Clear cell papillary renal cell carcinoma: a distinct histo-
pathologic and molecular genetic entity. Am J Surg Pathol 2008; 32: 1239. 14. Gobbo S, Eble JN, Maclennan GT et al: Renal cell carcinomas with papillary architecture and clear cell components: the utility of immunohistochemical and cytogenetical analyses in differential diagnosis. Am J Surg Pathol 2008; 32: 1780. 15. Edge SB, Byrd DR, Compton CC et al: AJCC Cancer Staging Manual, 7th ed. New York: Springer 2009. 16. Frank I, Blute ML, Cheville JC et al: An outcome prediction model for patients with clear cell renal cell carcinoma treated with radical nephrectomy based on tumor stage, size, grade and necrosis: the SSIGN score. J Urol 2002; 168: 2395. 17. Swets JA: Measuring the accuracy of diagnostic systems. Science 1988; 240: 1285. 18. Halat S, Eble JN, Grignon DJ et al: Multilocular cystic renal cell carcinoma is a subtype of clear cell renal cell carcinoma. Mod Pathol 2010; 23: 931.
Purpose: Cystic renal cell carcinoma has more favorable biology than noncystic renal cell carcinoma. Recently cystic change detected grossly or by low power microscopy was found to be a good prognostic factor for clear cell renal cell carcinoma. We assessed the optimal cutoff value of the proportion of cystic change with prognostic significance for clear cell renal cell carcinoma. Materials and Methods: We identified 223 patients with clear cell renal cell carcinoma who underwent partial or radical nephrectomy between 2001 and 2003. The cystic proportion of the tumor cut surface was calculated objectively and its prognostic significance was evaluated. Results: The ROC curve showed that a cystic percent of between 6% and 10% was appropriate to detect patients with renal cell carcinoma at low risk for cancer mortality and progression. A cutoff of 6% was adopted as a break point of cystic change for patient stratification. We analyzed the records of 87 patients (39.0%) with tumors with a cystic proportion of greater than 5%, that is 6% or greater. They had significantly lower stage and lower Fuhrman nuclear grade than patients with tumors with a cystic change of 5% or less (each p ⬍0.0001). On multivariate analysis a cystic proportion of more than 5% was a good prognostic indicator of cancer specific and progression-free survival (HR 0.221, p ⫽ 0.044 and HR 0.214, p ⫽ 0.004, respectively). Conclusions: In patients with clear cell renal cell carcinoma a cystic change comprising more than 5% of the tumor is a good independent predictor of survival.
Abbreviations and Acronyms ccRCC ⫽ clear cell RCC MCRCC ⫽ multilocular cystic RCC RCC ⫽ renal cell carcinoma Submitted for publication October 28, 2010. Study received Seoul National University Hospital institutional review board approval. Supported by the Basic Science Research Program through the National Research Foundation of Korea, funded by Ministry of Education, Science and Technology Grant 2010-0004550. * Correspondence: Department of Pathology and Kidney Research Institute, Medical Research Center, Seoul National University College of Medicine, 28 Chongno-gu Yeongeon-dong, Seoul 110-799, Republic of Korea (telephone: 82-2-20721767; FAX: 82-2-7432-5530; e-mail: blue7270@ snu.ac.kr).
Key Words: kidney; carcinoma, renal cell; cysts; mortality; Republic of Korea CYSTIC change is not rare in RCC.1 This feature was found in 4% to 15% of RCCs in an imaging study2 and in approximately 30% of ccRCCs examined in a recent histopathological review.3 RCC with a predominantly cystic pattern of growth, defined as neoplastic cells occupying 25% or less of tumor volume, was considered cystic RCC. Some groups have reported that cystic RCC has more favorable biology than noncystic RCC.3– 8 Currently
MCRCC, which is composed of an entirely cystic lesion, is recognized as a distinct subtype of ccRCC with excellent prognosis in the 2004 WHO classification of tumors.9 –11 Previously we reported that a cystic change detected on gross examination or low power microscopy was a good prognostic factor for ccRCC despite the exclusion of MCRCC.3 To our knowledge no group has systematically examined the association
0022-5347/11/1862-0423/0 THE JOURNAL OF UROLOGY® © 2011 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION
Vol. 186, 423-429, August 2011 Printed in U.S.A. DOI:10.1016/j.juro.2011.03.107
AND
RESEARCH, INC.
www.jurology.com
423
424
RENAL CELL CANCER WITH SIGNIFICANT CYSTIC CHANGE
between the proportion of cystic change and prognosis in patients with ccRCC. In this study we determined an optimal cystic proportion cutoff value to define 2 groups of patients with ccRCC with statistically significantly different outcomes.
MATERIALS AND METHODS Patient Selection We reviewed the pathology slides of consecutively resected ccRCCs between 2001 and 2003 at Seoul National University Hospital. MCRCC or RCC with features suggesting clear cell papillary RCC were excluded from study. MCRCC was defined according to 2004 WHO criteria as a tumor entirely composed of multilocular cysts lined by thin septa containing clear tumor cells with no expansile solid nodule on gross or microscopic examination.12 Clear cell papillary RCC is a distinctive group of renal tumors composed mainly of cells with clear cytoplasm arranged in cystic and papillary patterns. Clear cell papillary RCC is confined to the kidney with no recurrence or metastasis.13,14 Also excluded from study were ccRCC in patients with von Hippel-Lindau syndrome, dialysis associated cystic RCC and metachronous ccRCC. A total of 223 patients with ccRCC were identified for analysis. This study was approved by the Seoul National University Hospital institutional review board. All slides were reviewed by 2 pathologists.
Clinicopathological Evaluation The clinical features included patient age at surgery, gender, surgery type and clinical outcome. Followup was defined as the interval between the date of nephrectomy and the date of death or last followup. The pathological features evaluated were tumor size, TNM stage according to the 2009 American Joint Committee on Cancer,15 regional
lymph node involvement, distant metastasis, nuclear grade according to the Fuhrman system and predominant cyst fluid content. If no pathological confirmation of metastatic disease was performed, patients were assessed for clinical metastatic stage based on clinical examination and radiological studies. Tumor size was measured as the longest dimension. Nuclear grade was based on the highest grade tumor area identified. Cysts were considered dilated spaces filled with amorphous fluid or red blood cells lined by clear tumor cells with a dimension of at least 0.1 cm (fig. 1, A to C). However, cases in which the tumor contained cystic spaces containing necrotic tumor tissue were excluded from study since RCC with cystic necrosis carries a worse prognosis (fig. 1, E).16 A lumen filled with fluid in a hyalinized area was also considered a cyst. The predominant fluid content in the cysts was categorized as predominantly serous with amorphous eosinophilic fluid accounting for 60% or greater of the whole cystic content, predominantly bloody with red blood cells accounting for 60% or greater of the whole cystic content and mixed. Red blood cell extravasation between tumor cells was not considered a cyst filled with blood (fig. 1, D).
Cystic Proportion Determination Hematoxylin and eosin glass slides of tumors of each case were scanned using a Super Coolscan® 9000 ED film scanner. The digitalized slide images and digital gross pictures were used to quantify the area of the lumina in the cystic component and the total area of the tumor cut surface with ImageJ (fig. 2). The proportion of the cystic component was calculated by dividing the sum of cystic luminal areas by the whole area of the tumor cut surface. Tumors with a cystic component of less than 1% were considered noncystic RCC (0% cystic). There were 2 potential errors when determining the cystic proportion in this study. 1) Operator error could
Figure 1. RCC features. A, gross view of small RCC shows no apparent cysts. B, microscopy of same tumor reveals many microcysts containing predominantly serous fluid (scan power). C, RCC with predominantly bloody cysts. Reduced from ⫻40. D, RCC with extravasation of red blood cells. Reduced from ⫻100. E, RCC with cystic necrosis. Reduced from ⫻40.
RENAL CELL CANCER WITH SIGNIFICANT CYSTIC CHANGE
425
RESULTS
Figure 2. Measurement of cystic and total tumor areas. Luminal space inside each cyst (light blue outlines) and tumor were selected as regions of interests. Areas of regions were measured by pixel count. Cystic component of tumor in this section is 8%.
have increased the cystic proportion when more cystic areas of tumor were sectioned to generate the pathology slides. We believe that this error was minor since gross specimens and slides were subsequently reviewed. 2) There was potential measurement error. We found cystic proportion differences of 1% to 2% when the cystic portion of the same tumor image was measured by the same individual using ImageJ software. To overcome the measurement error we clustered RCC cases based on the percent of cystic area, including 0% (entirely solid), 1% to 5%, 6% to 10%, 11% to 15%, 16% to 20%, 21% to 25%, 26% to 30%, 31% to 35%, 36% to 40% and greater than 40%.
Statistical Analysis We calculated the area under the ROC curve according to the cystic proportion as a measure to predict a low risk of cancer specific death and progression. The optimal ROC curve threshold matched the cystic percent cutoff point showing improved sensitivity and specificity. An AUC of 0.5 to 0.7 indicated a low accuracy test, between 0.7 and 0.9 indicated moderate accuracy and more than 0.9 indicated high accuracy.17 Patients were classified into 2 groups by the identified optimal cutoff value. The Pearson chi-square and Student t tests were used to compare categorical and continuous variables, respectively. Cancer specific survival was measured from the date of surgery to the date of cancer related death or last followup. Progression-free survival was calculated from the date of surgery to the date of tumor recurrence/metastasis or last followup. Survival analysis was performed using the Kaplan-Meier method with results compared using the log rank test. Multivariate analysis was performed with Cox proportional hazards regression models. All statistical analysis was done using SPSS®, version 17.0 with p ⬍0.05 considered statistically significant.
Table 1 lists patient characteristics. Median age at surgery was 55 years (range 28 to 82) and 75.3% of the patients were male. Median followup was 64 months (range 2 to 97). Of the 223 patients 33 (14.8%) died of RCC and 12 (5.4%) died of another cause. A total of 164 patients (73.5%) had no evidence of RCC and 14 (6.3%) had disease progression. Median tumor size on pathological examination was 4 cm (range 1 to 16). In all tumors the size of the cyst was 0.1 to 4 cm. Cyst fluid contents were predominantly serous in 74 cases (50.0%), bloody in 49 (33.1%) and mixed in 25 (16.9%). The RCC cystic proportion was 0% to 80% (median ⫾ SD 9% ⫾ 14%). Of the tumors 88 (39.5%) were entirely solid, 48 (21.5%) had cysts taking up 1% to 5% of the tumor and 87 (39.0%) RCCs had a cystic proportion that accounted for more than 5% of the tumor. ROC curve analysis showed that the cystic component ratio had moderate accuracy to predict a low risk of cancer related death (mean AUC 0.742 ⫾ 0.042, 95% CI 0.660 – 0.823) and disease progression (AUC 0.745 ⫾ 0.037, 95% CI 0.673– 0.817). ROC curve data revealed that a cystic proportion of Table 1. Clinical and pathological characteristics of 223 patients with ccRCC Median ⫾ SD age (range) No. gender (%): F M Median ⫾ SD cm tumor size (range) No. nephrectomy (%): Partial Radical No. TNM stage (%): 1 2 3 4 No. Fuhrman nuclear grade (%): 1 2 3 4 No. predominant cystic content (%): Serous Bloody Mixed Median ⫾ SD % cystic component (range) No. cystic component (%): 0% 1%–5% 6%–10% 11%–15% 16%–20% 21%–25% 26%–30% 31%–35% 36%–40% Greater than 40%
55 ⫾ 11.2 (28–82) 55 (24.7) 168 (75.3) 4 ⫾ 3.1 (1–16) 32 191
(14.3) (85.7)
153 22 26 22
(68.6) (9.9) (11.6) (9.9)
22 102 76 23
(9.9) (45.7) (34.1) (10.3)
74 (50%) 49 (33.1%) 25 (16.9%) 9 ⫾ 14 (0–80) 88 48 23 15 9 10 7 10 3 10
(39.5) (21.5) (10.3) (6.7) (4.1) (4.5) (3.1) (4.5) (1.3) (4.5)
426
RENAL CELL CANCER WITH SIGNIFICANT CYSTIC CHANGE
Table 2. ROC curve data on sensitivity and specificity of each cystic cutoff No Ca Specific Death % Cystic Cutoff 1–5 6–10 11–15 16–20 21–25 26–30 31–35 36–40 41 or Greater
No Progression
% Sensitivity
% Specificity
Likelihood Ratio
% Sensitivity
% Specificity
Likelihood Ratio
66.8 44.7 32.6 24.7 20 15.3 11.6 6.8 5.3
75.8 93.9 93.9 93.9 93.9 97 97 100 100
2.76 7.33 5.34 4.05 3.28 5.1 3.87 . .
69.3 47.2 34.7 26.7 21.6 16.5 12.5 7.4 5.7
72.3 91.5 93.6 95.7 95.7 97.9 97.9 100 100
2.5 5.55 5.42 6.21 5.02 7.86 5.95 . .
between 6% and 10% was appropriate to identify patients with RCC at no risk for cancer specific death or disease progression (table 2). Sensitivity for no cancer related death and progression-free survival was 44.7% and 47.2% with 93.9% and 91.5% specificity, respectively. We adopted a cystic portion of 6% as the optimal cutoff to distinguish 2 groups of patients with RCC who had significantly different outcomes. A total of 136 patients (61.0%) had tumors with a cystic proportion of 5% or less while 87 (39.0%) had tumors with a cystic proportion of greater than 5% (table 3). RCCs with a cystic component of more than 5% of the tumor had lower stage and lower Fuhrman nuclear grade (each p ⬍0.0001). RCCs with cystic lesions of more than 5% of the tumor were associated with significantly longer cancer specific and progression-free survival than those with cystic lesions of 5% or less of the tumor (log rank p ⬍0.0001, fig. 3). On univariate analysis lower stage, lower Fuhrman nuclear grade, and a cystic proportion of more than 2% and 10% were also predictive variables of cancer related mortality and progression. However, the predominant cystic fluid content was not significantly related to patient outcome on survival analysis. Multivariate analysis revealed that a cystic proportion of greater than 5% (model 2) was an indeTable 3. Clinicopathological characteristics of patients with RCC by 5% cutoff % Cystic Proportion
No. pts (%) Median age No. gender (%): F M No. stage (%): I–II III–IV No. Fuhrman nuclear grade (%): 1–2 3–4
5 or Less
Greater Than 5
136 (61) 58
87 (39) 51
38 (27.9) 98 (72.1)
17 (19.5) 70 (80.5)
95 (69.9) 41 (30.1)
80 (92) 7 (8)
70 (51.5) 66 (48.5)
54 (62.1) 33 (37.9)
p Value 0.429 0.156
⬍0.0001 ⬍0.0001
pendent, good prognostic variable for cancer specific and progression-free survival (HR 0.221, 95% CI 0.051– 0.963, p ⫽ 0.044 and HR 0.214, 95% CI 0.054 – 0.613, p ⫽ 0.004, respectively), as were stage and Fuhrman nuclear grade (table 4). A cystic proportion of greater than 2% (model 1) and greater than 10% (model 3) were independent, good prognostic predictors of progression-free survival but they were not statistically significant for cancer specific survival on multivariate analysis. To determine whether cystic change contributes a survival benefit over tumor stage and nuclear grade we performed stratified survival analysis (fig. 4). A cystic proportion of more than 5% of tumor stratified by tumor stage was associated with better cancer specific and progression-free survival (p ⫽ 0.0140 and 0.0012, respectively, fig. 4, A and B). A cystic proportion of more than 5% of tumor stratified by nuclear grade also showed additional survival benefit for cancer specific and progression-free survival (p ⫽ 0.0001 and ⬍0.0001, respectively, fig. 4, C and D).
DISCUSSION Conventional RCC predominantly composed of multilocular cysts is considered cystic RCC. Cystic RCC reportedly accounts for 2.8% to 12% of RCCs and is associated with more favorable outcomes, according to some groups.4,7– 8 Since cystic RCC has been defined as a tumor with a cystic component representing at least 75% of the tumor,5–7 these studies may have included MCRCCs, which has a cystic component of almost 100%. MCRCC is currently defined by strict morphological criteria according to the 2004 WHO classification of tumors and no tumor with this feature has ever recurred or metastasized.12 MCRCC is differentiated from extensively cystic ccRCC by the absence of a grossly solid area or an expansile tumor cell mass on microscopy.18 In some studies the term cystic RCC includes MCRCC.4,8 Thus, the frequency of good outcomes of RCC with cystic change may possibly have been overestimated.
RENAL CELL CANCER WITH SIGNIFICANT CYSTIC CHANGE
427
Figure 3. Kaplan-Meier survival curves by cystic component with 5% cutoff for all patients (log rank test for trend p ⬍0.0001). A, cancer specific survival. B, progression-free survival.
We recently reported that a cystic change observed grossly or by low power microscopy was still statistically significantly associated with a lower risk of cancer progression in patients with ccRCC even when excluding those with MCRCC.3 Since MCRCC carries a good prognosis and RCC with cystic change also has a better prognosis than solid RCC, it may be hypothesized that patient survival is proportional to the percent of the cystic component and a cutoff point separates groups with significantly different prognoses. To our knowledge we report the first study of the association between the proportion of cystic change in ccRCC and clinical outcome. In previous studies the cystic component was calculated by subtracting the solid component from
total tumor area.5–7 This method does not reflect real tumor cell-free volume since neoplastic cells exist around the cysts. In this study to specifically select the cancer-free area, we defined the cystic component as the area inside the cyst. Since assessment of the cystic proportion by microscopy alone was not reproducible, pixel counts of the area of the cyst lumina and of the total tumor mass in the same digitalized image were measured using ImageJ software and the percent of the cystic component was calculated using those values. The ROC curve determined that cysts were a moderate predictor of a good clinical outcome in patients and a cystic proportion of between 6% and 10% was adequate to identify patients with greater
Table 4. Multivariate models by cystic proportion cutoff Proportion Hazards Model (prognostic variables) Model 1: Stage Nuclear grade % Cystic proportion Model 2: Stage Nuclear grade % Cystic proportion Model 3: Stage Nuclear grade % Cystic proportion Model 4: Stage Nuclear grade % Cystic proportion
Ca Specific Survival Category
Progression-Free Survival
HR (95% CI)
p Value
HR (95% CI)
p Value
cutoff
I–II vs III–IV 1–2 vs 3–4 2 or Less vs greater than 2
12.862 (4.914–33.667) 4.611 (1.540–13.805) 0.451 (0.177–1.149)
⬍0.0001 0.006 0.095
7.358 (3.743–14.467) 4.478 (2.007–9.989) 0.328 (0.154–0.700)
⬍0.0001 ⬍0.0001 0.004
cutoff
I–II vs III–IV 1–2 vs 3–4 5 or Less vs greater than 5
11.63 (4.467–30.284) 4.2 (1.390–12.691) 0.221 (0.051–0.963)
⬍0.0001 0.011 0.044
7.129 (3.637–13.976) 4.181 (1.865–9.371) 0.214 (0.074–0.613)
⬍0.0001 0.001 0.004
cutoff
I–II vs III–IV 1–2 vs 3–4 10 or Less vs greater than 10
14.059 (5.479–36.077) 4.072 (1.350–12.282) 0.359 (0.083–1.556)
⬍0.0001 0.013 0.171
8.353 (4.298–16.234) 3.879 (1.735–8.669) 0.263 (0.080–0.864)
⬍0.0001 0.001 0.028
cutoff
I–II vs III–IV 1–2 vs 3–4 20 or Less vs greater than 20
15.99 (6.341–40.317) 4.29 (1.433–12.846) 0.546 (0.128–2.324)
⬍0.0001 0.009 0.413
9.864 (5.034–18.628) 3.867 (1.743–8.579) 0.293 (0.070–1.221)
⬍0.0001 0.001 0.092
428
RENAL CELL CANCER WITH SIGNIFICANT CYSTIC CHANGE
Figure 4. Stratified Kaplan-Meier survival curves. A, cancer specific survival adjusted for stage (integrated over strata log rank test p ⫽ 0.0140). B, progression-free survival adjusted for stage (integrated over strata log rank test, p ⫽ 0.0012). C, cancer specific survival adjusted for nuclear grade (integrated over strata log rank test p ⫽ 0.0001). D, progression-free survival adjusted for nuclear grade (integrated over strata log rank test p ⬍0.0001).
than 90% specificity and with a high likelihood ratio. A cutoff of 6% was adopted as the break point of cystic change to stratify cases. RCC with a cystic proportion of greater than 5% was significantly associated with lower stage and lower nuclear grade than RCC with a cystic proportion of 5% or less. This result implies that tumors with a cystic change accounting for a relatively small portion of the whole tumor nodule also have favorable biology. Univariate analysis revealed that cystic change based on the 2%, 5% and 10% cutoffs was a statistically significant, good prognostic predictor of cancer specific and progression-free survival. However, only cystic change according to the 5% cutoff significantly decreased the risk of cancer specific death and disease progression on multivariate analysis. Significant survival differences between RCCs with a greater than 5% cystic lesion and RCCs with a cystic lesion of 5% or less were also seen in subgroups after stratifying patients into favorable (stages I–II and grades 1–2) and unfavorable (stages III–IV and grades 3– 4) groups. Thus, a cystic proportion of more than 5% is a favorable predictive factor regardless of stage or nuclear grade in ccRCC cases. The p values for correlations between the cystic percent and outcomes on univariate and multivariate analyses tended to increase as the cutoff increased. This tendency could have been the result
of grouping cases with a moderate amount of cystic area, such as 15%, but less than the cutoff, such as 20%, in the no cystic change group, although they had a good prognosis. Additional studies in large populations with much longer followup are required to clarify the relationship between cystic change and cancer prognosis. Cysts are occasionally seen during gross examination of the ccRCC specimen. Sometimes cysts are only apparent after microscopic examination due to obscuring cyst fluid or small cyst size. The presence of cysts is not enough to predict survival until they make up 5% of the tumor. Careful gross and microscopic examinations are needed to detect ccRCC with clinically significant cysts.
CONCLUSIONS Cystic change in ccRCC can be described by dilated cystic spaces with a diameter of greater than 0.1 cm regardless of cystic fluid content, except for necrotic tumor tissue. A cystic change of more than 5% of the tumor is an independent, good prognostic factor in patients with ccRCC. Careful pathological examination to detect ccRCC with clinically significant cystic change is needed. Future studies are needed to ascertain the pathogenesis of cystic change.
RENAL CELL CANCER WITH SIGNIFICANT CYSTIC CHANGE
429
REFERENCES 1. Grignon DJ and Che M: Clear cell renal cell carcinoma. Clin Lab Med 2005; 25: 305. 2. Hartman DS, Davis CJ Jr, Johns T et al: Cystic renal cell carcinoma. Urology 1986; 28: 145. 3. Park HS, Jung EJ, Myung JK et al: The prognostic implications of cystic change in clear cell renal cell carcinoma. Korean J Pathol 2010; 44: 149. 4. Bielsa O, Lloreta J and Gelabert-Mas A: Cystic renal cell carcinoma: pathological features, survival and implications for treatment. BJU 1998; 82: 16. 5. Corica FA, Iczkowski KA, Cheng L et al: Cystic renal cell carcinoma is cured by resection: a study of 24 cases with long-term followup. J Urol 1999; 161: 408. 6. Nassir A, Jollimore J, Gupta R et al: Multilocular cystic renal cell carcinoma: a series of 12 cases and review of the literature. Urology 2002; 60: 421. 7. Han KR, Janzen NK, McWhorter VC et al: Cystic renal cell carcinoma: biology and clinical behavior. Urol Oncol 2004; 22: 410.
8. Webster W, Thompson R, Cheville J et al: Surgical resection provides excellent outcomes for patients with cystic clear cell renal cell carcinoma. Urology 2007; 70: 900. 9. Imura J, Ichikawa K, Takeda J et al: Multilocular cystic renal cell carcinoma: a clinicopathological, immuno- and lectin histochemical study of nine cases. APMIS 2004; 112: 183. 10. Suzigan S, Lopez-Beltran A, Montironi R et al: Multilocular cystic renal cell carcinoma: a report of 45 cases of a kidney tumor of low malignant potential. Am J Clin Pathol 2006; 125: 217. 11. Gong K, Zhang N, He Z et al: Multilocular cystic renal cell carcinoma: an experience of clinical management for 31 cases. J Cancer Res Clin Oncol 2008; 134: 433. 12. Lopez-Beltran A, Scarpelli M, Montironi R et al: 2004 WHO classification of the renal tumors of the adults. Eur Urol 2006; 49: 798. 13. Gobbo S, Eble JN, Grignon DJ et al: Clear cell papillary renal cell carcinoma: a distinct histo-
pathologic and molecular genetic entity. Am J Surg Pathol 2008; 32: 1239. 14. Gobbo S, Eble JN, Maclennan GT et al: Renal cell carcinomas with papillary architecture and clear cell components: the utility of immunohistochemical and cytogenetical analyses in differential diagnosis. Am J Surg Pathol 2008; 32: 1780. 15. Edge SB, Byrd DR, Compton CC et al: AJCC Cancer Staging Manual, 7th ed. New York: Springer 2009. 16. Frank I, Blute ML, Cheville JC et al: An outcome prediction model for patients with clear cell renal cell carcinoma treated with radical nephrectomy based on tumor stage, size, grade and necrosis: the SSIGN score. J Urol 2002; 168: 2395. 17. Swets JA: Measuring the accuracy of diagnostic systems. Science 1988; 240: 1285. 18. Halat S, Eble JN, Grignon DJ et al: Multilocular cystic renal cell carcinoma is a subtype of clear cell renal cell carcinoma. Mod Pathol 2010; 23: 931.