- Email: [email protected]
Analysis of imaging modalities, staging systems, and prognostic indicators for renal cell carcinoma
ANALYSIS OF IMAGING MODALITIES, STAGING SYSTEMS, AND PROGNOSTIC INDICATORS FOR RENAL CELL CARCINOMA S. G. LANNON, M.D. F. G. MACK, M.D. 0. H. MILLARD, M.D.
C. I? N. DINNEY, M.D. S. A. AWAD, M.D. J. B. GAJEWSKI, M.D. I? BELITSKY, M.D.
From the Department of Urology, Dalhousie University and Affiliated Teaching Hospitals, Halifax, Nova Scotia, Canada ABSTRACT-A retrospective analysis of314 patients with renal cell carcinoma was donefocusing mainly on imaging modalities and prognostic significance of tumor stage using both the Robson and TNM systems. Computerized tomography (CT) scan proved to be the most effective modality for staging. Overall staging accuracy was 62 and 68 percent for TNM and Robson staging, respectively, and understaging was more frequent than overstaging. The actuarial five-year survival using the Robson system was 73 percent for Stage A, 68 percent Stage B, 51 percent Stage C, and 20 percent Stage D. The main limitation of the Robson system is the heterogeneity of the Stage C group which includes patients with renal vein and those with nodal involvement with a significant difference in survival. The survival by the TNM system showed no difference in those with Tl, T2, T3a and T3b disease but a significant difference in those with T3c or T4a. One hundred sixteen patients (37%) presented with metastatic disease with a median survival of seventeen months (range 2-204) for those with solitary metastasis and six months (range 1-132) for those with multiple metastases (the difference was not statistically significant). Except for anecdotal cases, nephrectomy with or without treatment of the metastases did not seem to affect survival significantly. The presence of spindle cell, alone or in association with clear or granular cell, affected the prognosis adversely. Thirtyone patients had their tumors identified incidentally. Their stage at diagnosis was earlier than the symptomatic group (Stage Tl-T2: 77% vs 34 %), and there was a significant difference in the disease-free survival at fifty-four months between the two groups (79% vs 57%, respectively).
While the treatment of renal cell carcinoma has not changed dramatically in recent years, new insight into the available diagnostic methods and prognostic factors have surfaced in the literature. Uncertainty as to the ideal staging system has also become apparent. With the general consensus that surgery is the only effective form of treatment for renal cell carcinoma, it is implicit that accurate clinical staging is necessary to identify patients who may benefit from surgery.‘m4 The main focus of our study
122
was to evaluate various imaging modalities in staging and to compare the accuracy of clinical staging by the Robson and TNM5 methods in relation to prognosis. Material
and Methods
We did a retrospective analysis of 314 patients treated at our teaching hospitals who presented between 1970-1986 with renal cell carcinoma. Included were 201 men with an
UROLOGY
/
FEBRUARY 1992
/
VOLUME XXXIX,
NUMBER 2
TABIX
Presenting symptoms
I.
SYMPTOM
______
NO
______
?? hemxturia
TABLE II.
(314 patients) ___
Staging values (Cohen/Kappa different imaging modalitie,y *
% ~_
111
35
85
27
*mass
24
8
?? paraneoplastic
54
17
syndrome ?? metastasis
38
12
Retroperitoneum
*others
30
IO
Colon
6
2
*flank
pain
*unknown
ANG
CT Size
< 5 cm
Size
’ 5 cm
U’S
test) of
IVP
INVASION Capsular
Duodenum IVC
*incidental
31
IO
Renal
vein
METASTASES Liver
age of 62.7 + 10.7 (range 30-82), and 113 women with an average age of 62.0 _+ 12.6 (range 26-80). One hundred fifty-three patients had a tumor on the right side, 150 patients had a tumor on the left, and 5 patients had bilateral tumors. In the remaining patients this information was not available. The tumor was discovered incidentally during abdominal ultrasound in 31 patients. The clinical presentation is shown. in Table I. Of the 314 patients 251 had renal surgery. Except for a few (28 patients) who had a simple nephrectomy in the early part of the series, most (206 patients) had a radical nephrectomy for the primary lesion. Partial nephrectomy (or enucleation) was done in 7 patients with single kidney or bilateral disease. In 10 patients the operative and pathology records were not available. Patients (63) who presented with a characteristic mass within the kidney and metastases compatible with renal cell carcinoma were included in the study even if they did not have renal surgery (i.e., no pathologic diagnosis). Clinicopathologic correlation of the primary lesion was done on patients who had renal surgery according to both the Robson and TNM!’ staging systems. We staged all patients using both Robson and the TNM classification systems to determine which was the best prognostic indicator. Clinical staging with CT scan, angiography, ultrasound, and intravenous pyelogram (IVP) was used to determine tumor size, capsular invasion, involvement of the renal vein and inferior vena cava (.IVC), invasion of the Gerota fascia average
UROLOCI
FI’RRUAHY
1992
/ \‘OLUME XXXIX,
Adrenal Lymph
nodes
*The Kappa values excellent
ri
good
(3,
and retroperitoneum, presence of lymph node, adrenal, and liver metastases, and invasion of the colon or duodenum. During the years covered by the study, CT scan supplanted angiography as the primary imaging modality. Ultrasound was often done prior to or in addition to CT scan to evaluate involvement of the IVC. CT scanning was performed using a GE 9800 Quick Scanner or the Phillips tomoscan 310, taking 9 mm cuts from the diaphragm to the iliac crest. Ultrasound examinations were done with commercially available gray scale static and real time scanners. Angiography was performed, by the Seldinger technique with an aortic flush, followed by selective Ncatheterization of the renal artery. The contralateral kidney was examined selectively. A cavogram was done in selected cases. Lung metastases were identified by chest x-ray film, lung tomography, or CT scan. Bone metastases were evaluated by either bone scan or in the earlier ye,ars by skeletal survey. The study examined tumor size a.nd cell type to determine their effect on survival. Information on cell type was available on 229 patients. In the 31 patients in whom the tumor was discovered incidentally, the pathologic stage, cell type, and survival were compared with those patients who presented with symptoms.
NUMRER
2
123
Clinical and pathologic correlation of Robson staging*
TABLE III.
Total
Pathological
Clinical
A
TABLE
A
0
C
D
106
7
14
23
8
14
2
0
24
C
4
3
23
5
35
*Patients with known metastases or local invasion (clinical Stage D) are not included in table.
STAQE
T
A L
1
2
3A
38
3C
30
4A
48
0
-
1
1
,
_
_
_
_
_
3
1
-
12
2
2
-
-
-
-
-
16
2
1
7
113
24
25
1
-
5
-
176
3A
-
-
4
6
2
-
-
1
-
13
38
-
-
1
1
6
1
-
1
-
12
3c
-
-
1
-
-
6
-
2
1
10
3D
-
_
_
_
_
_
_
_
_
0
PA
-
-
2
1
2
-
-
5
1
11
4S
-
_
_
_
_
_
_
_
_
0
Correlation between clinical and pathologic Stage N
TABLE V.
STAGE C L I N Ic A L
N
0
1
2
0
211
12
7
1
3
3
2
8
2
1
0
2
3
TABLE VI.
76
95
82
T2
77
92
82
T3a
84
65
58
T3b
51
65
31
T3c
30
65
T4a
19
analysis
Survival analysis was based on the KaplanMeier method6 with p 0.05 considered to be the level of significance. The Z-test7 was used when the data were insufficient to apply the KaplanMeier method. The Student t-test and chisquared test were used when indicated. The Cohen/Kappa tests was used to measure the relationship between clinical and pathologic staging. Kappa values greater than 0.70 may be taken to represent excellent agreement beyond chance, values between 0.40 and 0.70 may be taken to represent fair to good agreement, and values below 0.40 or so may be taken to represent poor agreement. Results
TOTAL
PATHOLOGICAL
SIMINOVITCH
Tl
Statistical TOTAL
PATHOLOQICAL 0
c L I N ‘c
Correlation between clinical and pathologic Stage T
IV.
SELLI
PRESENT SERIES
150
B
TABLE
STAGE
Crude jive-year survival (%) by TNM System4,20
VII.
230
Crude jive-year survival ( %) by Robson staging3J6,1s
Imaging
modalities
CT scan, angiography, ultrasound, and IVP were compared for their accuracy in defining local characteristics of renal tumors (Table II). The Kappa values, according to the Cohen/ Kappa test, were graded excellent, fair to good, and poor. Applying this gradation, the CT scan, on balance, is shown to be vastly superior to the other imaging modalities. In the five categories where CT scan was graded poor, all other imaging modalities were graded poor as well. The superior value of the CT scan was noted particularly in tumors less than 5 cm. Staging
STAGE
PRESENT SERIES
ROBSON
GOLIMBU
KONNAK
A
73
66
88
65
B
68
64
67
50
C
51
42
40
57
D
20
11
2
0
124
Correlation between clinical and pathologic stage for Robson is shown in Table III and for TNM in Tables IV and V. Excluding patients with distant metastases, 68 percent of the patients were accurately staged, 24 percent were understaged, and 7 percent were overstaged by the Robson system. By the TNM system the T stage of the tumor was accurate in 62 percent,
UROLOGY
i
FEBRUARY
1992
i
VOLUME XXXIX.
NUMBER 2
TAHLE:
-___
VIII.
Site and distribution
of metastases SURVIVAL
SITE -_____
NO
%
%
. -.-m.mm.--,
,
,
_
I
-
LUNG
52
45
BONE
39
34
LIVER
20
17
ADRENAL
8
7
BRAIN
7
6
SKIN
5
4
EYE
1
1
I
O-
OTHER
23
0
5 SURVIVAL
20
10 TIME IN YEARS
15
20
____ STAGE
understaged in 30 percent, and overstaged in 8 percent (Table IV). The nodal status was accurately staged in 90 percent and understaged in 8 percent (Table V). We combined all of our modalities in determining staging accuracy and found th.at understaging (30 % ) was a more common error than overstaging (8 %). Survival
by stage o.f the disease
Survival distribution by the Robson pathologic staging is shown in Figure 1. The estimated five-year survival rate was 73 percent for Stage A, 68 percent for Stage B, 51 percent for Stage C, and 20 percent for Stage D. Statistically there was a highly significant difference in survival between the various stages (p = 0.001). The survival curve for the pathologic T stage is shown in Figure 2. We found no significant differences in survival among Stages Tl, T2, T3a, and T3b using Kaplan-Meier (p = 0.18). However there was a highly significant difference lin survival between these stages and Stages T3c and T4a, as well as between these two stages (p = 0.0001). A comparison between our five-year survivals and those of other series is shown in Table VI (Robson system) and Table VII (TNM system). There was an apparent difference in survival between NOM0 and NlMO (Fig. 3). However, there were too few patients in the NlMO stage for statistical analysis. Figure 4 shows a highly significant difference between the survival curves for MO and Ml (I? = 0.0001).
-A
FIGURE1. C, and D.
i-0
*C
OD
Survival curves for Robson Stages
B,
A,
100 80 60 T2
40 20
’
T3a
.i
0 0
5 SURVIVAL ___
I
10 TIME IN YEARS ~_~ ~~~~ _~
16
20
STAGE
--
FIGURE
Tl
-
T2
T3c
-
148
2.
T3a
.
Life
.~
T3b
exp.
Survival curves for Stages T,I--4a (NO,
MO). SURVIVAL
%
IO
40
NIMO 20
;Metasta tic disease One hundred sixteen patients (37%) presented with metastatic disease, 31 patients with a solitary metastasis, and 85 patients had multiple metastases. The site and frequency of all metastases are shown in Table VIII.
STAGE 1 -NO,
FIGURE:3.
MO
+
Nl, MO
Survival curves for Stages NO and Nl
CM@.
125
TABLE IX.
I
CELL TYPE
20
STAGE +
Ml
*
Life exp.
I
Survival curves for Stages MO and Ml.
FIGURE4.
1
60 -
40 -1
20 -
0
J
._.
_______.._
0
._~
5
10
~.
5~- ----~~ .r .~~.- .-,
15 YEARS
20
25
30
Metastases -
--F, Multiple
Solitary
x
Life exp.
I
Survival curves for patients with solitary and multiple metastases.
FIGURE 5.
% SURVIVE
o___
__
__~__~,__
0
__.~~~~.
~,
!
CELL -
CLEAR
i -*-
CLEAR
d GRAN.
FICLJHE6. cell types.
126
.-
-
-.-
10 YEARS
5
~_
-----~-J
15
20
TYPE
.-
GRANULAR
*
CL L GR d SP
Survival curvesfor
*
SPINDLE
15
7
6
3
CLEAR & GRANULAR
27
12
CLEAR & GRANULAR % SPINDLE
11
5
3
1
SPINDLE
I j
% 72
GRANULAR
I
NO 165
CLEAR
;
--MO
Histologic cell type
OTHERS
Median survival for 31 patients with a single metastasis was seventeen months (range Z-204 months) and for those with multiple metastases, six months (range 1-132 months). Analysis by the Kaplan-Meier method failed to show a significant difference (Fig. 5). Twenty-one patients with a solitary metastasis underwent nephrectomy. Median survival for this group was twenty-one months (range 4141 months). For those who did not undergo nephrectomy median survival was ten months (range 2-204 months). The difference was not statistically significant. We compared survival between patients who had nephrectomy plus surgical excision or radiotherapy to the solitary metastasis, with those who had either nephrectomy only or no treatment at all. While the numbers are small, there may be improved surviva1 for those patients with a solitary bone metastasis when both nephrectomy and treatment of the solitary metastasis were carried out (median survival 46 months). Four of the 9 patients are currentIy clinically free of disease following nephrectomy and treatment of a solitary metastasis (range 33-141 months), 1 patient had a solitary lung metastasis, 2 patients had a solitary bone metastasis, and 1 patient had a solitary metastasis in the left supraclavicular lymph node. Of 85 patients with multiple metastases, 32 underwent nephrectomy. Median survival was eight months (range 2-132 months). Those patients who did not undergo nephrectomy had a median survival of six months (range 1-175 months). There was no significant difference. No spontaneous remissions occurred in any of the 53 patients with solitary or multiple metastases who underwent nephrectomy. Cell type
patients according to
The distribution of cell types is shown in Table IX. The Kaplan-Meier survival, by cell
UHOI,OGY
/
FEBRUARY 1RYZ : VOLlJMEXXXIX,NUMBEB 2
‘rAHI,l<
-______-
Stage at diagnosis
x.
STAGE
INCIDENTAL -_
SYMPTOMATIC _____
Tl-T2
24 (779/o)*
96
(34%)*
T3-T4
3
(10%)
60
(22%)
Nl Ml
4
(13%b
121
(44%b
TOTAL 31 ____ ‘Chi-squaw tct. p
NO OF PATIENTS .~ ~~ 30
-INCIDENTAL
m
SYMPTOMAl-IC
277
TAHIX XI.
Disease-free survival at mean follow-up (54 month,s)
____
71
72
73
74
75
76
77
78
79
80
81
62
83
84
85
86
YEARS
INCIDENTAL
SYMPTOMATIC
Tl-T2
19 (79%)*
55
(57%b
T3-T4
2 (67%)
24
(40%)
STAGE
70
__-
F1c1.w i. incidence o_f incidental matic tumor.s during study period.
and syrnf)to-
f 40
type, is shown in Figure 6. There was a significant survival advantage for patients with granular cell tumors or those with combined clear and granular cells over those with spindle cells (p = 0.0057). lncident,al
:
F
:
:
20 ~*
0, 0
5
10 YEARS
15
20
DIAGNOSIS
tumors
Thirty-one patients had their tumors identified incidentally while being investigated for an unrelateld or nonspecific complaint. Although there wals an increase in incidence of incidental tumors during the study period, actual proportion to s,ymptomatic tumors remains the same (Fig. 7). There was no significant difference in age or sex distribution or cell type between the two groups. Table X demonstrates that incidental tumors presented at an earlier stage with a greater crude survival (Fig. 8) (p = 0.004). When disease-free survival was compared between similar stages of the two groups, there was a significant difference in survival for Stage TIT2 (Table XI). Comment Imaging
i
modalities
Most studies in the literature demonstrated that CT scan is the optimal imaging modality1.Y while angiography has been less accurate for
-
INCIDENTAL
FICUHP: 8. Survival symptomatic tflm0r.s.
curves
.
jar
SYMPTOMATIC
incidental
I
and
staging.“.“’ LJsing angiography Bracken and Jonsson’” were able to accuratel\r stage their patients by Robson stage in 35 percent, understaged in 32 percent and overstaged in 33 percent. Bassil, Dosoretz, and Prout” using TNM and various imaging modalities but not CT scanning found staging was accurate in 48 percent, understaged in 47 percent, and a’verstaged in 5 percent. Frohmuller, Grups, and Heller2 using the TNM system found accurate staging in 72 percent with CT scan, 78 percent with ultrasound, S7 percent with angiography, and 59 percent with IVP, They found that CT scan tended to overstage tumors (19 % ). while ultrasound (18 % ), angiography (29 % ). and IVP (33%) more often understaged them. On balance, these reports support our conclusion. It
becomes apparent that there is no indication to routinely perform angiography when staging renal tumors. The use of angiography should be limited to patients in whom conservative surgery is contemplated, where the tumor is very large to identify the arterial supply, or when the diagnosis is questionable. Staging Staging is simpler by the Robson classification than the TNM system. However, a real limitation of using the Robson system lies in the Stage C which includes patients with renal vein, inferior vena cava, and nodal involvement. Our estimated five-year survival for patients with renal vein involvement is 60 percent, compared with 40 percent for those with nodal disease and 25 percent for those with supradiaphragmatic inferior vena cava involvement. This group is clearly heterogeneous and for prognostic purposes should be stratified. The defect of TNM is in its complexity” which makes it difficult to apply and the benefit at times is not apparent. For example, there was no difference in survival between Tl, T2, T3a, and T3b disease. The TNM system is also being further refined, for example, the classification T3 is used to identify tumors that involve the perinephric fat (T3a) and the subdiaphragmatic inferior vena cava (T3c). As a general observation, the use of Robson and TNM interchangeably makes it more difficult to analyze the aggregate results of renal carcinoma staging. Perhaps a common staging system is needed that will combine the advantage of both Robson and TNM systems. Metastatic disease Several studies suggested a significant difference in survival between patients with single metastasis and those with multiple metastases.12-13Except for anecdotal cases, other studies did not show a distinct pattern of prolonged survival or spontaneous regression in patients with solitary metastasis. 14,15We were not able to demonstrate a difference between patients with single and multiple metastases probably because of the wide range in the survivor period in both groups. Opinion is divided as to whether or not aggressive surgery is conducive to patient survival. Some authors favor aggressive surgery, while others found no evidence for its efficacy in prolonging survival. Our results except for few rare cases emphasize the poor prognosis for patients with metastatic disease, whether or not
128
surgery is performed. As in so many instances, the decision to operate except for symptoms is subjective. Cell type Our results are consistent with those in the literature insofar as they reflect better survival with clear or granular as compared with spindle or anaplastic cell type. Opinions differ as to whether clear cell or granular cell indicates a Some studies reveal no difbetter prognosis. ferencere while other studies indicate a survival advantage for clear cells.r7 The difference in our series was not statistically significant. Incidental tumors These patients represented a much smaller proportion of patients (10%) in our series than has been observed in other studies. 18~r9Konnack and Grossman1a found that the incidence of incidental tumors has increased in recent years making up 48 percent of 46 patients treated between 1980 and 1984. While we have seen an increase in the number of new cases presenting incidentally over the past ten years, we have noted a parallel increase in the total number of cases of renal cell carcinoma seen in the same period. The proportion of incidental tumors does not appear to have changed. Our data are in agreement with the literature, that incidental tumors have a better prognosis because they present at a lower stage. We also found that for Stages Tl-T2, there was better survival for incidental tumors. The reason for this difference should be further explored. The cost benefit of a mass screening program to detect asymptomatic renal cell carcinoma remains to be determined. Conclusion Over a broad range of issues, our study tends to confirm past findings. In particular we conclude that CT scan is by far the most effective imaging modality for staging renal cell carcinoma. We also conclude that the spindle cell type is predictive of poor survival. At perhaps more substantive level, there is a need to continue to address the traditional staging systems and devise a new system that is practical, relates to the most reliable diagnostic modalities, and has prognostic significance.
UROLOGY
Victoria General Hospital Halifax, Nova Scotia Canada B3H 2Y9 (DR. AWAD)
/ FEBRUARY
1992
i
VOLUME
XXXIX, NUMBER 2
References 1. Lang EK: Comparison of dynamic and conventional computed to:mography, angiography, and ultrasonography in the staging of renal cell carcinoma, Cancer 54: 2205 (1984). 2. Frohmuller HCW, Grups JW, and Heller V: Comparative value of ultrasonography, computerized tomography, angiography and excretory urography in the staging of renal cell carcinoma, J Ural 138: 482 (1987). 3. Robson CJ, Churchill BM, and Anderson W: The results of radical ncphrectomy for renal cell carcinoma, J Ural 101: 297 (1969). 4. Selii C, Hinshaw WM, Woodard BH, and Paulson DF: Stratification of risk factors in renal cell carcinoma, Cancer 52: 899 (1983). 5. Harmer MI-I: TNM Classification of Malignant Tumours, 3rd ed, Geneva, International Union Against Cancer, 1978. 6. Kaplan EL, and Meier P: Nonparametric estimation from incomphete observations J Am Stat Assoc 53: 457 (1958). 7. Huntsberger DV, and Billingsley P: Elements of Statistical Inference, Allyn and Bacon, Inc., 5th Ed, 1981, p 392. 8. Fleiss JL: Statistical Methods for Rates and Proportions, John Wiley & Sons, 2nd Ed, 1981, p 218. 9. Cronan JJ, Zeman RK, and Rosenfield AT: Comparison of computerized tomography, ultrasound and angiography in staging renal cell carcinoma, J Urol 127: 712 (1982). 10. Bracken B, and Jonsson K: How accurate is angiographic
UROLOCY
FEBRUARY
1992
/ VOLUME
XXXIX, NUMBER
staging of renal carcinoma? Urology 14: 96 (1979). 11. Bassil B, Dosoretz DE, and Prom CR: Validatiorr of the tumor, nodes and metastasis, classification 01 renal cell carcinoma, J Ural 134: 450 (1985). 12. Dineen MDK, Pastore RD, Emrich I,J, and Huhen HP: Results of surgical treatment of renal cell carcinoma with solitary metastasis, J Urol 140: 277 (1988). 13. Neves RJ, Zincke H. and Taylor WY: Metastatic rtnal cell cancer and radical nephrectomy: identification of ~mqmostic fartors and patient survival, J Urol 139: 1173 (1988:~. 14. Johnson DE, Kaesier KE, and Samuc& MI>: Is nephrectomy justified with metastatic renal cell carcmoma’?, J Ural 114: 27 (1975). 15. Garfield DH, and Kennedy BJ: Progression of mctastatic renal cell carcinoma following nrphrectoru!: Cancer 29: 190 (1972). 16. Golimbu M. Al-Askari S, Tessler A. and Morales P: Aggressive treatment of metastatic renal cancer, J Ural 1.36: 805 (1986). 17. Reis M, and Faria V: Renal carcinoma. !Rwvaluation of prognostic factors, Cancer 61: 1192 (1988). 18. Konnak JW, and Grossman IIB: Renal cell 1:arcinoma as an incidental finding, J Urol 134: 1094 (1985). 19. Thompson IM, and Peek M: Improvement in survival ol patients with renal cell carcinoma-the role of the wrmdipitously detected tumor. J Urol 140: 487 (1988). 20. Siminovitch JMP, Montie JE, Straffon RA: Prognostic indicators in renal adenocarcinoma, J Ural 130: 20 (1983).
2
129
C. I? N. DINNEY, M.D. S. A. AWAD, M.D. J. B. GAJEWSKI, M.D. I? BELITSKY, M.D.
From the Department of Urology, Dalhousie University and Affiliated Teaching Hospitals, Halifax, Nova Scotia, Canada ABSTRACT-A retrospective analysis of314 patients with renal cell carcinoma was donefocusing mainly on imaging modalities and prognostic significance of tumor stage using both the Robson and TNM systems. Computerized tomography (CT) scan proved to be the most effective modality for staging. Overall staging accuracy was 62 and 68 percent for TNM and Robson staging, respectively, and understaging was more frequent than overstaging. The actuarial five-year survival using the Robson system was 73 percent for Stage A, 68 percent Stage B, 51 percent Stage C, and 20 percent Stage D. The main limitation of the Robson system is the heterogeneity of the Stage C group which includes patients with renal vein and those with nodal involvement with a significant difference in survival. The survival by the TNM system showed no difference in those with Tl, T2, T3a and T3b disease but a significant difference in those with T3c or T4a. One hundred sixteen patients (37%) presented with metastatic disease with a median survival of seventeen months (range 2-204) for those with solitary metastasis and six months (range 1-132) for those with multiple metastases (the difference was not statistically significant). Except for anecdotal cases, nephrectomy with or without treatment of the metastases did not seem to affect survival significantly. The presence of spindle cell, alone or in association with clear or granular cell, affected the prognosis adversely. Thirtyone patients had their tumors identified incidentally. Their stage at diagnosis was earlier than the symptomatic group (Stage Tl-T2: 77% vs 34 %), and there was a significant difference in the disease-free survival at fifty-four months between the two groups (79% vs 57%, respectively).
While the treatment of renal cell carcinoma has not changed dramatically in recent years, new insight into the available diagnostic methods and prognostic factors have surfaced in the literature. Uncertainty as to the ideal staging system has also become apparent. With the general consensus that surgery is the only effective form of treatment for renal cell carcinoma, it is implicit that accurate clinical staging is necessary to identify patients who may benefit from surgery.‘m4 The main focus of our study
122
was to evaluate various imaging modalities in staging and to compare the accuracy of clinical staging by the Robson and TNM5 methods in relation to prognosis. Material
and Methods
We did a retrospective analysis of 314 patients treated at our teaching hospitals who presented between 1970-1986 with renal cell carcinoma. Included were 201 men with an
UROLOGY
/
FEBRUARY 1992
/
VOLUME XXXIX,
NUMBER 2
TABIX
Presenting symptoms
I.
SYMPTOM
______
NO
______
?? hemxturia
TABLE II.
(314 patients) ___
Staging values (Cohen/Kappa different imaging modalitie,y *
% ~_
111
35
85
27
*mass
24
8
?? paraneoplastic
54
17
syndrome ?? metastasis
38
12
Retroperitoneum
*others
30
IO
Colon
6
2
*flank
pain
*unknown
ANG
CT Size
< 5 cm
Size
’ 5 cm
U’S
test) of
IVP
INVASION Capsular
Duodenum IVC
*incidental
31
IO
Renal
vein
METASTASES Liver
age of 62.7 + 10.7 (range 30-82), and 113 women with an average age of 62.0 _+ 12.6 (range 26-80). One hundred fifty-three patients had a tumor on the right side, 150 patients had a tumor on the left, and 5 patients had bilateral tumors. In the remaining patients this information was not available. The tumor was discovered incidentally during abdominal ultrasound in 31 patients. The clinical presentation is shown. in Table I. Of the 314 patients 251 had renal surgery. Except for a few (28 patients) who had a simple nephrectomy in the early part of the series, most (206 patients) had a radical nephrectomy for the primary lesion. Partial nephrectomy (or enucleation) was done in 7 patients with single kidney or bilateral disease. In 10 patients the operative and pathology records were not available. Patients (63) who presented with a characteristic mass within the kidney and metastases compatible with renal cell carcinoma were included in the study even if they did not have renal surgery (i.e., no pathologic diagnosis). Clinicopathologic correlation of the primary lesion was done on patients who had renal surgery according to both the Robson and TNM!’ staging systems. We staged all patients using both Robson and the TNM classification systems to determine which was the best prognostic indicator. Clinical staging with CT scan, angiography, ultrasound, and intravenous pyelogram (IVP) was used to determine tumor size, capsular invasion, involvement of the renal vein and inferior vena cava (.IVC), invasion of the Gerota fascia average
UROLOCI
FI’RRUAHY
1992
/ \‘OLUME XXXIX,
Adrenal Lymph
nodes
*The Kappa values excellent
ri
good
(3,
and retroperitoneum, presence of lymph node, adrenal, and liver metastases, and invasion of the colon or duodenum. During the years covered by the study, CT scan supplanted angiography as the primary imaging modality. Ultrasound was often done prior to or in addition to CT scan to evaluate involvement of the IVC. CT scanning was performed using a GE 9800 Quick Scanner or the Phillips tomoscan 310, taking 9 mm cuts from the diaphragm to the iliac crest. Ultrasound examinations were done with commercially available gray scale static and real time scanners. Angiography was performed, by the Seldinger technique with an aortic flush, followed by selective Ncatheterization of the renal artery. The contralateral kidney was examined selectively. A cavogram was done in selected cases. Lung metastases were identified by chest x-ray film, lung tomography, or CT scan. Bone metastases were evaluated by either bone scan or in the earlier ye,ars by skeletal survey. The study examined tumor size a.nd cell type to determine their effect on survival. Information on cell type was available on 229 patients. In the 31 patients in whom the tumor was discovered incidentally, the pathologic stage, cell type, and survival were compared with those patients who presented with symptoms.
NUMRER
2
123
Clinical and pathologic correlation of Robson staging*
TABLE III.
Total
Pathological
Clinical
A
TABLE
A
0
C
D
106
7
14
23
8
14
2
0
24
C
4
3
23
5
35
*Patients with known metastases or local invasion (clinical Stage D) are not included in table.
STAQE
T
A L
1
2
3A
38
3C
30
4A
48
0
-
1
1
,
_
_
_
_
_
3
1
-
12
2
2
-
-
-
-
-
16
2
1
7
113
24
25
1
-
5
-
176
3A
-
-
4
6
2
-
-
1
-
13
38
-
-
1
1
6
1
-
1
-
12
3c
-
-
1
-
-
6
-
2
1
10
3D
-
_
_
_
_
_
_
_
_
0
PA
-
-
2
1
2
-
-
5
1
11
4S
-
_
_
_
_
_
_
_
_
0
Correlation between clinical and pathologic Stage N
TABLE V.
STAGE C L I N Ic A L
N
0
1
2
0
211
12
7
1
3
3
2
8
2
1
0
2
3
TABLE VI.
76
95
82
T2
77
92
82
T3a
84
65
58
T3b
51
65
31
T3c
30
65
T4a
19
analysis
Survival analysis was based on the KaplanMeier method6 with p 0.05 considered to be the level of significance. The Z-test7 was used when the data were insufficient to apply the KaplanMeier method. The Student t-test and chisquared test were used when indicated. The Cohen/Kappa tests was used to measure the relationship between clinical and pathologic staging. Kappa values greater than 0.70 may be taken to represent excellent agreement beyond chance, values between 0.40 and 0.70 may be taken to represent fair to good agreement, and values below 0.40 or so may be taken to represent poor agreement. Results
TOTAL
PATHOLOGICAL
SIMINOVITCH
Tl
Statistical TOTAL
PATHOLOQICAL 0
c L I N ‘c
Correlation between clinical and pathologic Stage T
IV.
SELLI
PRESENT SERIES
150
B
TABLE
STAGE
Crude jive-year survival (%) by TNM System4,20
VII.
230
Crude jive-year survival ( %) by Robson staging3J6,1s
Imaging
modalities
CT scan, angiography, ultrasound, and IVP were compared for their accuracy in defining local characteristics of renal tumors (Table II). The Kappa values, according to the Cohen/ Kappa test, were graded excellent, fair to good, and poor. Applying this gradation, the CT scan, on balance, is shown to be vastly superior to the other imaging modalities. In the five categories where CT scan was graded poor, all other imaging modalities were graded poor as well. The superior value of the CT scan was noted particularly in tumors less than 5 cm. Staging
STAGE
PRESENT SERIES
ROBSON
GOLIMBU
KONNAK
A
73
66
88
65
B
68
64
67
50
C
51
42
40
57
D
20
11
2
0
124
Correlation between clinical and pathologic stage for Robson is shown in Table III and for TNM in Tables IV and V. Excluding patients with distant metastases, 68 percent of the patients were accurately staged, 24 percent were understaged, and 7 percent were overstaged by the Robson system. By the TNM system the T stage of the tumor was accurate in 62 percent,
UROLOGY
i
FEBRUARY
1992
i
VOLUME XXXIX.
NUMBER 2
TAHLE:
-___
VIII.
Site and distribution
of metastases SURVIVAL
SITE -_____
NO
%
%
. -.-m.mm.--,
,
,
_
I
-
LUNG
52
45
BONE
39
34
LIVER
20
17
ADRENAL
8
7
BRAIN
7
6
SKIN
5
4
EYE
1
1
I
O-
OTHER
23
0
5 SURVIVAL
20
10 TIME IN YEARS
15
20
____ STAGE
understaged in 30 percent, and overstaged in 8 percent (Table IV). The nodal status was accurately staged in 90 percent and understaged in 8 percent (Table V). We combined all of our modalities in determining staging accuracy and found th.at understaging (30 % ) was a more common error than overstaging (8 %). Survival
by stage o.f the disease
Survival distribution by the Robson pathologic staging is shown in Figure 1. The estimated five-year survival rate was 73 percent for Stage A, 68 percent for Stage B, 51 percent for Stage C, and 20 percent for Stage D. Statistically there was a highly significant difference in survival between the various stages (p = 0.001). The survival curve for the pathologic T stage is shown in Figure 2. We found no significant differences in survival among Stages Tl, T2, T3a, and T3b using Kaplan-Meier (p = 0.18). However there was a highly significant difference lin survival between these stages and Stages T3c and T4a, as well as between these two stages (p = 0.0001). A comparison between our five-year survivals and those of other series is shown in Table VI (Robson system) and Table VII (TNM system). There was an apparent difference in survival between NOM0 and NlMO (Fig. 3). However, there were too few patients in the NlMO stage for statistical analysis. Figure 4 shows a highly significant difference between the survival curves for MO and Ml (I? = 0.0001).
-A
FIGURE1. C, and D.
i-0
*C
OD
Survival curves for Robson Stages
B,
A,
100 80 60 T2
40 20
’
T3a
.i
0 0
5 SURVIVAL ___
I
10 TIME IN YEARS ~_~ ~~~~ _~
16
20
STAGE
--
FIGURE
Tl
-
T2
T3c
-
148
2.
T3a
.
Life
.~
T3b
exp.
Survival curves for Stages T,I--4a (NO,
MO). SURVIVAL
%
IO
40
NIMO 20
;Metasta tic disease One hundred sixteen patients (37%) presented with metastatic disease, 31 patients with a solitary metastasis, and 85 patients had multiple metastases. The site and frequency of all metastases are shown in Table VIII.
STAGE 1 -NO,
FIGURE:3.
MO
+
Nl, MO
Survival curves for Stages NO and Nl
CM@.
125
TABLE IX.
I
CELL TYPE
20
STAGE +
Ml
*
Life exp.
I
Survival curves for Stages MO and Ml.
FIGURE4.
1
60 -
40 -1
20 -
0
J
._.
_______.._
0
._~
5
10
~.
5~- ----~~ .r .~~.- .-,
15 YEARS
20
25
30
Metastases -
--F, Multiple
Solitary
x
Life exp.
I
Survival curves for patients with solitary and multiple metastases.
FIGURE 5.
% SURVIVE
o___
__
__~__~,__
0
__.~~~~.
~,
!
CELL -
CLEAR
i -*-
CLEAR
d GRAN.
FICLJHE6. cell types.
126
.-
-
-.-
10 YEARS
5
~_
-----~-J
15
20
TYPE
.-
GRANULAR
*
CL L GR d SP
Survival curvesfor
*
SPINDLE
15
7
6
3
CLEAR & GRANULAR
27
12
CLEAR & GRANULAR % SPINDLE
11
5
3
1
SPINDLE
I j
% 72
GRANULAR
I
NO 165
CLEAR
;
--MO
Histologic cell type
OTHERS
Median survival for 31 patients with a single metastasis was seventeen months (range Z-204 months) and for those with multiple metastases, six months (range 1-132 months). Analysis by the Kaplan-Meier method failed to show a significant difference (Fig. 5). Twenty-one patients with a solitary metastasis underwent nephrectomy. Median survival for this group was twenty-one months (range 4141 months). For those who did not undergo nephrectomy median survival was ten months (range 2-204 months). The difference was not statistically significant. We compared survival between patients who had nephrectomy plus surgical excision or radiotherapy to the solitary metastasis, with those who had either nephrectomy only or no treatment at all. While the numbers are small, there may be improved surviva1 for those patients with a solitary bone metastasis when both nephrectomy and treatment of the solitary metastasis were carried out (median survival 46 months). Four of the 9 patients are currentIy clinically free of disease following nephrectomy and treatment of a solitary metastasis (range 33-141 months), 1 patient had a solitary lung metastasis, 2 patients had a solitary bone metastasis, and 1 patient had a solitary metastasis in the left supraclavicular lymph node. Of 85 patients with multiple metastases, 32 underwent nephrectomy. Median survival was eight months (range 2-132 months). Those patients who did not undergo nephrectomy had a median survival of six months (range 1-175 months). There was no significant difference. No spontaneous remissions occurred in any of the 53 patients with solitary or multiple metastases who underwent nephrectomy. Cell type
patients according to
The distribution of cell types is shown in Table IX. The Kaplan-Meier survival, by cell
UHOI,OGY
/
FEBRUARY 1RYZ : VOLlJMEXXXIX,NUMBEB 2
‘rAHI,l<
-______-
Stage at diagnosis
x.
STAGE
INCIDENTAL -_
SYMPTOMATIC _____
Tl-T2
24 (779/o)*
96
(34%)*
T3-T4
3
(10%)
60
(22%)
Nl Ml
4
(13%b
121
(44%b
TOTAL 31 ____ ‘Chi-squaw tct. p
NO OF PATIENTS .~ ~~ 30
-INCIDENTAL
m
SYMPTOMAl-IC
277
TAHIX XI.
Disease-free survival at mean follow-up (54 month,s)
____
71
72
73
74
75
76
77
78
79
80
81
62
83
84
85
86
YEARS
INCIDENTAL
SYMPTOMATIC
Tl-T2
19 (79%)*
55
(57%b
T3-T4
2 (67%)
24
(40%)
STAGE
70
__-
F1c1.w i. incidence o_f incidental matic tumor.s during study period.
and syrnf)to-
f 40
type, is shown in Figure 6. There was a significant survival advantage for patients with granular cell tumors or those with combined clear and granular cells over those with spindle cells (p = 0.0057). lncident,al
:
F
:
:
20 ~*
0, 0
5
10 YEARS
15
20
DIAGNOSIS
tumors
Thirty-one patients had their tumors identified incidentally while being investigated for an unrelateld or nonspecific complaint. Although there wals an increase in incidence of incidental tumors during the study period, actual proportion to s,ymptomatic tumors remains the same (Fig. 7). There was no significant difference in age or sex distribution or cell type between the two groups. Table X demonstrates that incidental tumors presented at an earlier stage with a greater crude survival (Fig. 8) (p = 0.004). When disease-free survival was compared between similar stages of the two groups, there was a significant difference in survival for Stage TIT2 (Table XI). Comment Imaging
i
modalities
Most studies in the literature demonstrated that CT scan is the optimal imaging modality1.Y while angiography has been less accurate for
-
INCIDENTAL
FICUHP: 8. Survival symptomatic tflm0r.s.
curves
.
jar
SYMPTOMATIC
incidental
I
and
staging.“.“’ LJsing angiography Bracken and Jonsson’” were able to accuratel\r stage their patients by Robson stage in 35 percent, understaged in 32 percent and overstaged in 33 percent. Bassil, Dosoretz, and Prout” using TNM and various imaging modalities but not CT scanning found staging was accurate in 48 percent, understaged in 47 percent, and a’verstaged in 5 percent. Frohmuller, Grups, and Heller2 using the TNM system found accurate staging in 72 percent with CT scan, 78 percent with ultrasound, S7 percent with angiography, and 59 percent with IVP, They found that CT scan tended to overstage tumors (19 % ). while ultrasound (18 % ), angiography (29 % ). and IVP (33%) more often understaged them. On balance, these reports support our conclusion. It
becomes apparent that there is no indication to routinely perform angiography when staging renal tumors. The use of angiography should be limited to patients in whom conservative surgery is contemplated, where the tumor is very large to identify the arterial supply, or when the diagnosis is questionable. Staging Staging is simpler by the Robson classification than the TNM system. However, a real limitation of using the Robson system lies in the Stage C which includes patients with renal vein, inferior vena cava, and nodal involvement. Our estimated five-year survival for patients with renal vein involvement is 60 percent, compared with 40 percent for those with nodal disease and 25 percent for those with supradiaphragmatic inferior vena cava involvement. This group is clearly heterogeneous and for prognostic purposes should be stratified. The defect of TNM is in its complexity” which makes it difficult to apply and the benefit at times is not apparent. For example, there was no difference in survival between Tl, T2, T3a, and T3b disease. The TNM system is also being further refined, for example, the classification T3 is used to identify tumors that involve the perinephric fat (T3a) and the subdiaphragmatic inferior vena cava (T3c). As a general observation, the use of Robson and TNM interchangeably makes it more difficult to analyze the aggregate results of renal carcinoma staging. Perhaps a common staging system is needed that will combine the advantage of both Robson and TNM systems. Metastatic disease Several studies suggested a significant difference in survival between patients with single metastasis and those with multiple metastases.12-13Except for anecdotal cases, other studies did not show a distinct pattern of prolonged survival or spontaneous regression in patients with solitary metastasis. 14,15We were not able to demonstrate a difference between patients with single and multiple metastases probably because of the wide range in the survivor period in both groups. Opinion is divided as to whether or not aggressive surgery is conducive to patient survival. Some authors favor aggressive surgery, while others found no evidence for its efficacy in prolonging survival. Our results except for few rare cases emphasize the poor prognosis for patients with metastatic disease, whether or not
128
surgery is performed. As in so many instances, the decision to operate except for symptoms is subjective. Cell type Our results are consistent with those in the literature insofar as they reflect better survival with clear or granular as compared with spindle or anaplastic cell type. Opinions differ as to whether clear cell or granular cell indicates a Some studies reveal no difbetter prognosis. ferencere while other studies indicate a survival advantage for clear cells.r7 The difference in our series was not statistically significant. Incidental tumors These patients represented a much smaller proportion of patients (10%) in our series than has been observed in other studies. 18~r9Konnack and Grossman1a found that the incidence of incidental tumors has increased in recent years making up 48 percent of 46 patients treated between 1980 and 1984. While we have seen an increase in the number of new cases presenting incidentally over the past ten years, we have noted a parallel increase in the total number of cases of renal cell carcinoma seen in the same period. The proportion of incidental tumors does not appear to have changed. Our data are in agreement with the literature, that incidental tumors have a better prognosis because they present at a lower stage. We also found that for Stages Tl-T2, there was better survival for incidental tumors. The reason for this difference should be further explored. The cost benefit of a mass screening program to detect asymptomatic renal cell carcinoma remains to be determined. Conclusion Over a broad range of issues, our study tends to confirm past findings. In particular we conclude that CT scan is by far the most effective imaging modality for staging renal cell carcinoma. We also conclude that the spindle cell type is predictive of poor survival. At perhaps more substantive level, there is a need to continue to address the traditional staging systems and devise a new system that is practical, relates to the most reliable diagnostic modalities, and has prognostic significance.
UROLOGY
Victoria General Hospital Halifax, Nova Scotia Canada B3H 2Y9 (DR. AWAD)
/ FEBRUARY
1992
i
VOLUME
XXXIX, NUMBER 2
References 1. Lang EK: Comparison of dynamic and conventional computed to:mography, angiography, and ultrasonography in the staging of renal cell carcinoma, Cancer 54: 2205 (1984). 2. Frohmuller HCW, Grups JW, and Heller V: Comparative value of ultrasonography, computerized tomography, angiography and excretory urography in the staging of renal cell carcinoma, J Ural 138: 482 (1987). 3. Robson CJ, Churchill BM, and Anderson W: The results of radical ncphrectomy for renal cell carcinoma, J Ural 101: 297 (1969). 4. Selii C, Hinshaw WM, Woodard BH, and Paulson DF: Stratification of risk factors in renal cell carcinoma, Cancer 52: 899 (1983). 5. Harmer MI-I: TNM Classification of Malignant Tumours, 3rd ed, Geneva, International Union Against Cancer, 1978. 6. Kaplan EL, and Meier P: Nonparametric estimation from incomphete observations J Am Stat Assoc 53: 457 (1958). 7. Huntsberger DV, and Billingsley P: Elements of Statistical Inference, Allyn and Bacon, Inc., 5th Ed, 1981, p 392. 8. Fleiss JL: Statistical Methods for Rates and Proportions, John Wiley & Sons, 2nd Ed, 1981, p 218. 9. Cronan JJ, Zeman RK, and Rosenfield AT: Comparison of computerized tomography, ultrasound and angiography in staging renal cell carcinoma, J Urol 127: 712 (1982). 10. Bracken B, and Jonsson K: How accurate is angiographic
UROLOCY
FEBRUARY
1992
/ VOLUME
XXXIX, NUMBER
staging of renal carcinoma? Urology 14: 96 (1979). 11. Bassil B, Dosoretz DE, and Prom CR: Validatiorr of the tumor, nodes and metastasis, classification 01 renal cell carcinoma, J Ural 134: 450 (1985). 12. Dineen MDK, Pastore RD, Emrich I,J, and Huhen HP: Results of surgical treatment of renal cell carcinoma with solitary metastasis, J Urol 140: 277 (1988). 13. Neves RJ, Zincke H. and Taylor WY: Metastatic rtnal cell cancer and radical nephrectomy: identification of ~mqmostic fartors and patient survival, J Urol 139: 1173 (1988:~. 14. Johnson DE, Kaesier KE, and Samuc& MI>: Is nephrectomy justified with metastatic renal cell carcmoma’?, J Ural 114: 27 (1975). 15. Garfield DH, and Kennedy BJ: Progression of mctastatic renal cell carcinoma following nrphrectoru!: Cancer 29: 190 (1972). 16. Golimbu M. Al-Askari S, Tessler A. and Morales P: Aggressive treatment of metastatic renal cancer, J Ural 1.36: 805 (1986). 17. Reis M, and Faria V: Renal carcinoma. !Rwvaluation of prognostic factors, Cancer 61: 1192 (1988). 18. Konnak JW, and Grossman IIB: Renal cell 1:arcinoma as an incidental finding, J Urol 134: 1094 (1985). 19. Thompson IM, and Peek M: Improvement in survival ol patients with renal cell carcinoma-the role of the wrmdipitously detected tumor. J Urol 140: 487 (1988). 20. Siminovitch JMP, Montie JE, Straffon RA: Prognostic indicators in renal adenocarcinoma, J Ural 130: 20 (1983).
2
129