Histologic predictors of renal cell carcinoma response to interleukin-2-based therapy.

Abstract

The authors examined pathology from patients with renal cancer (RCC) treated with IL-2 to determine response rates for clear cell and variant RCC and to identify histologic features that predict response. Pathology specimens were reviewed by a single pathologist who was blinded to both the prior pathology interpretation and the therapeutic response. Findings were correlated with response to IL-2 therapy. Evaluable pathology specimens were obtained from 231 patients. Of 163 primary RCCs, the response rate was 21% (30/146) for patients with clear cell versus 6% (1/17) for patients with variant or indeterminate type RCC (P = 0.20). For clear cell carcinomas, response to IL-2 was associated with the presence of alveolar features and the absence of papillary and granular features. Patients with more than 50% alveolar features and no granular or papillary features had a 39% response rate (14/36). Patients with alveolar and granular features representing less than 50% of the specimen and no papillary features had a 19% response rate (15/77). The response rate for the others was 3% (1/33). This model was then applied to an independent sample of 68 metastasis specimens. Response rates in the three prognostic groups and for patients with non-clear cell cancers were 25% (5/20), 9% (2/22), 0% (0/16), and 0% (0/10), respectively. Median survivals for all patients with clear cell tumors by risk group were 2.87, 1.36, and 0.87 years, respectively (P < 0.001). These data suggest that patients with non-clear cell RCC or with clear cell RCC with papillary, no alveolar, and/or more than 50% granular features respond poorly to IL-2 and should be considered for alternative treatments. Investigation of other tumor-related predictors of IL-2 responsiveness is warranted.

Full text

Page 1

CLINICAL STUDY
Histologic Predictors of Renal Cell Carcinoma Response
to Interleukin-2-Based Therapy
Melissa P. Upton,* Robert A. Parker,† Amanda Youmans,‡ David F . McDermott,‡
and Michael B. Atkins‡
Summary: The authors examined pathology from patients with
renal cancer (RCC) treated with IL-2 to determine response rates for
clear cell and variant RCC and to identify histologic features that
predict response. Pathology specimens were reviewed by a single
pathologist who was blinded to both the prior pathology interpretation
and the therapeutic response. Findings were correlated with response
to IL-2 therapy. Evaluable pathology specimens were obtained from
231 patients. Of 163 primary RCCs, the response rate was 21%
(30/146) for patients with clear cell versus 6% (1/17) for patients with
variant or indeterminate type RCC (P = 0.20). For clear cell
carcinomas, response to IL-2 was associated with the presence of
alveolar features and the absence of papillary and granular features.
Patients with more than 50% alveolar features and no granular or
papillary features had a 39% response rate (14/36). Patients with
alveolar and granular features representing less than 50% of the
specimen and no papillary features had a 19% response rate (15/77).
The response rate for the others was 3% (1/33). This model was then
applied to an independent sample of 68 metastasis specimens.
Response rates in the three prognostic groups and for patients with
non-clear cell cancers were 25% (5/20), 9% (2/22), 0% (0/16), and
0% (0/10), respectively. Median survivals for all patients with clear
cell tumors by risk group were 2.87, 1.36, and 0.87 years,respectively
(P , 0.001). These data suggest that patientswith non-clear cell RCC
or with clear cell RCC with papillary, no alveolar, and/or more than
50% granular features respond poorly to IL-2 and should be
considered for alternative treatments. Investigation of other tumor-
related predictors of IL-2 responsiveness is warranted.
Key Words: renal cancer, IL-2, histologic features
(J Immunother 2005;28:488–495)
M
approved therapy, produces tumor regression in 15% to 20%
of patients,1,2with only 7% of patients showing durable long-
term freedom from recurrence. High-dose IL-2 treatment is also
associated with considerable toxicity and expense, making it an
impractical standard. Improved treatment strategies and/or better
methods of identifying patients likely to benefit from IL-2-based
therapy are needed.
Kidney carcinomas have been shown to comprise several
distinct subtypes that differ in genetics as well as in histologic
appearance and clinical behavior.3,4The majority of conventional
(clear cell) carcinomas possess mutations in the Von Hippel-
Lindau gene located at chromosome sub-band 3p25, while
papillary (chromophil) carcinomas have alterations in copy
numbers, typically trisomies of chromosomes 7 and 17, and
losses of chromosome Y.3,5–7In addition, type I and type II
papillary carcinomas differ in their chromosomal abnormalities,
the latter more commonly showing losses in chromosome Xp,8
while many type I and familial papillary carcinomas have
mutations in the c-met oncogene (locus 7q31), which encodes
for hepatocyte growth factor receptor.9Chromophobe and
collecting duct carcinomas presumably have other associated
genetic abnormalities yet to be fully defined. The relative re-
sponsiveness of these various histologic subgroups to IL-2-based
therapy has never been fully evaluated.
Studies of RCC populations have identified clinical and
histologic factors that predict for poor survivalwith or without
specific therapy.10–24For example, Motzer et al reported that in
patients receiving IFN, poor survival is associated with low
Karnofsky performance status, high serum lactate dehydro-
genase, low hemoglobin, high ‘‘corrected’’serum calcium, and
time from initial RCC diagnosis to start of therapy of less than
1 year.21Similar clinical prognostic features have been identified
in patients receiving various IL-2-based treatments.22–24A re-
cent multivariate analysis confined to patients who received
IL-2 after nephrectomy reported survival to be inversely asso-
ciated with lymph node involvement, constitutional symptoms,
metastases involving sites other than bone or lung or multiple
sites, a TSH level more than 2.0 mIU/L, or sarcomatoid
histology.22Othershavereportedhighnucleargradeandrhabdoid
and sarcomatoid phenotypes to be associated with more ag-
gressive disease.14,15,25–27Prior to the distinction of papillary,
chromophobe, and collecting duct tumor types from conven-
tional carcinomas, morphologic features such as patterns of
cell arrangement, (tubular, papillary, and alveolar), degree
of granularity or eosinophilia of the cytoplasm, and extent
etastatic renal cell carcinoma (RCC) remains a vexing
therapeutic challenge. High-dose IL-2, the only FDA-
Received for publication July 23, 2004; accepted March 31, 2005.
From the *Beth Israel Deaconess Medical Center, Department of Pathology;
†Biometrics Center, and ‡Division of Hematology/Oncology, Harvard
Medical School, Boston, Massachusetts.
Supported by a grant from the Kidney Cancer Association, Evanston, IL.
M.P.U. is currently at the Anatomic Pathology Division, Department of
Pathology, University of Washington Medical Center, Seattle, WA. R.A.P.
is currently at the Center for Biostatistics in AIDS Research, Harvard
School of Public Health, Boston, MA.
Reprints: Michael B. Atkins, MD, Division of Hematology/Oncology,
Department of Medicine, Beth Israel Deaconess Medical Center, 330
Brookline Avenue, Boston, MA 02215 (e-mail: matkins@bidmc.harvard.
edu).
Copyright ? 2005 by Lippincott Williams & Wilkins
488
J Immunother?Volume 28, Number 5, September/October 2005

Page 2

of inflammation and/or necrosis were determined to have possi-
ble prognostic significance.11–17Worse survival was associated
with granular rather than clear cytoplasm and fusiform or
spindle-shaped cells.
Less information has been reported on predictors of
response to specific therapies, including IL-2. Clinical factors
that have beenvariably associated with response to IL-2-based
therapy include good performance status, one metastatic site,
no bone metastases or prior IFN therapy, prior nephrectomy or
erythropoietin production, treatment-related thrombocytope-
nia, no thyroid dysfunction, or rebound lymphocytosis, and
post-treatment elevations of blood TNF-a and IL-1 levels.28,29
Cangiona et al reported that despite a poor overall prognosis,
patients with sarcomatoid RCC were still ableto respond,30but
little additional information is available regarding the relation-
ship of other pathologic and morphologic features to tumor
response in patients receiving IL-2-based therapy.
Between 1990 and 2001, the Cytokine Working Group
(CWG) performed seven separate clinical trials involving
388 patientswith metastatic RCC using avarietyof IL-2-based
regimens.31These trials have included patients with variant
RCCs. Tumor responses were seen in 63 patients (19 complete
and 44 partial responses) for a response rate of 16%. An
additional 56 patients were treated with IL-2 at Beth Israel
Deaconess Medical Center (BIDMC) outside of these CWG
trials, with tumor responses observed in 12% of patients. We
performed a detailed histologic review and classification of
pathology slides obtained from these patients to determine the
response to IL-2 for patients with clear cell or variant RCC and
within in the clear cell subtype to identify morphologic features
predictive of response to IL-2-based therapy.
METHODS
Pathology slides were requested from all 388 patients
treated between 1990 and 2001 on CWG clinical trials. All
patients had previously provided informed consent to participate
in the CWG clinical trial, which included the possibility of
central review of their pathology material. Pathology speci-
mens were also obtained from patients treated on non-CWG
clinical trials involving IL-2 at BIDMC between 1997 and 2001.
The current investigation was approved by the Institutional
Review Board of BIDMC.
The CWG trials for which pathology specimens were
collected are described elsewhere.31Eligibility criteria for
these trials included histologically confirmed stage IV RCC,
measurable and clearly progressive disease, Eastern Cooperative
Oncology Group performance status of 0 or 1, excellent organ
function, and no prior IL-2-based therapy. Treatment included
inpatient regimens involving high-dose IL-2 alone or in com-
bination with IFNa or low-dose IL-2 alone or outpatient re-
gimensinvolvingIL-2incombinationwithIFNa withorwithout
5-flurouracil.31,32
Two hundred sixty-five specimens were obtained from
239 patients. Specimens were obtained from 183 patients on
CWG trials and all patients on BIDMC trials. Eleven specimens
from eight patients were eliminated from the analysis either
because of insufficient material (n = 4) or because they were
fine-needle aspirates (n = 7). In addition, 21 patients had two
specimens and 1 patient had three specimens. In these instances,
kidney and then lung metastasis specimens were preferentially
analyzed. Other metastatic sites were analyzed only when neither
kidney nor lung specimens were available. Only a single spec-
imen per patient was included in the analysis, so that any rule
developed for the kidney specimens could be assessed in an
independent group of patients with specimens obtained from
the other sites. Thus, we report results based on 231 pathology
specimens, from 231 distinct patients.
CaseswereclassifiedasIL-2respondersor non-responders
for this analysis, based on the response reported in the initial
clinical study. Responders included patients with either complete
or partial responses (.50% regression of measurable tumor
deposits). Patients were considered non-responders if there
was only a minor response (25%–50% regression of measur-
able tumor deposits) or no response to IL-2 treatment. Survival
was calculated from the date of initiation of IL-2 treatment
until date of either death or last follow-up.
Histologic evaluation was performed using standard
H&E-stained sections and specially stained slides available
from the original pathologic examination. All cases were eval-
uated according to a protocol based on the consensus clas-
sification of carcinoma types4and the UICC staging and
Fuhrman nuclear grading systems,10with additional data fields
designed to capture morphologic features that might permit
additional analyses. Semiquantitative assessments of necrosis
and tumor-infiltrating lymphocytes (TILs) were performed
using an Olympus BH-2 microscope with standard 103 oculars
and objectives.
All pathology assessments were performed by the same
pathologist (M.P.U.), blinded to the response to therapy and to
the original reading of the slides. The type of carcinoma was
recorded as clear cell, which may have variable amounts of
clear or granular/eosinophilic components; papillary with
subtyping of type I (basophilic) and type II (eosinophilic) as
defined by Delahunt and Eble33; chromophobe; collecting duct;
indeterminate; or other. The designation ‘‘indeterminate’’ was
reserved for cases having insufficient well-preserved tissue to
permit accurate classification, or for neoplasms so poorly dif-
ferentiated that assignment by type was not possible.
Each case was also assessed for features such as sarco-
matoid, rhabdoid, papillary, granular/eosinophilic, tubular, alve-
olar, and cystic morphologies, and it was noted whether or not
these features represented greater than 50% of the carcinoma
area. The architectural patterns of ‘‘tubular,’’ ‘‘alveolar,’’ and
‘‘cystic’’ growth arrangement were taken from the categori-
zation outlined and illustrated by Bostwick and Eble.34The
tubular pattern is also described as ‘‘acinar’’ byother authors.35
Standard definitions were used for sarcomatoid,27rhabdoid,25,26
and granular patterns. These patterns are illustrated in Figure 1.
The presence or absence of TILs and necrosis was noted
for each neoplasm. TIL presence was graded in a semiquantita-
tive fashion on a scale of 0 to 3, with 0 = no lymphocytes
identified; 1 = lymphocytes apparent in only one high-power
field (4003) or visible only at high power; 2 = lymphocytes
present on scanning power in some but not all areas of the
carcinoma and intermediate in concentration; 3 = abundant
lymphocytes seen at 1003 in many fields and in high con-
centration in at least two high-power fields. Necrosis was also
q 2005 Lippincott Williams & Wilkins
489
J Immunother?Volume 28, Number 5, September/October 2005 Histologic Predictors of RCC Response to IL-2

Page 3

graded 0 to 3, with 0 = no necrosis identified, 1 = necrosis
present in less than one high-power field (4003) and present in
less than two areas; 2 = necrosis present at scanning power
(1003) in less than two areas and occupying less than the area
of one high-power field; 3 = necrosis present with confluence
involving more than the area of one high-power field in at least
two areas of the slide.
The Fuhrman nuclear grades I to IV were assessed for
each carcinoma using standard definitions.10For each carcinoma
the percentage area with each nuclear grade was estimated and
recorded as less than 5%, 5% to 25%, 25% to 50%, or greater
than 50%, with additional scoring of the predominant grade
and the overall grade. The predominant grade was considered
thegradeoccupyingthegreatestpercentageareaoftheneoplasm.
The overallgradewas assigned as originally noted byFuhrman
et al10as ‘‘the most malignant or highest grade exhibited even
if only focal.’’
Finally, all available staging data, including invasion
of perinephric tissues, involvement of renal veins, speci-
fying whether major, minor, or both types of veins were
involved, and involvement of regional lymph nodes, were
recorded.
Data Analysis
The primary analysis examined the association of path-
ologic features and response to IL-2. Given that the majority of
cases had kidneyspecimens available, we developed predictive
rules using data only from primary kidney carcinomas. Because
over 90% of the kidney neoplasms were clear cell carcinomas,
and because we observed only one response in a patient with
non-clear cell carcinoma, we developed a predictive rule for
clear cell carcinomas only. We screened each variable for an
association with response, using either Fisher exact test (for
a binary variable) or Mantel-Haenszel test for trend when
there were ordered categories, with P , 0.20 used to identify
potential associations between response and each variable in
our database. Two-tailed P values, unadjusted for multiple com-
parisons, were used. Coding of ordered categorical variables
was optimized to permit maximum separation of response rate.
For example, we compared the degree of separation between
FIGURE 1. Examples of RCC conven-
tional variant (clear cell type) exhib-
iting (A and B) alveolar morphology:
the cells form a solid nested pattern
bounded by a connective tissue sep-
tum showing prominent vascularity
in an arborizing pattern that is del-
icate, branching, and reminiscent of
the alveolar septae of the lung; C and
D, tubular morphology: cells line the
connective tissue septum, with a lu-
men-like space in the center; A and
C, clear cell features: cells have
a foamy cytoplasm; B and D, gran-
ular cell features: cell cytoplasm has
an eosinophilic and finely granular
appearance; E, rhabdoid features:
cells are large and epithelioid cells
withvesicularnuclei,
nucleoli, and prominent paranuclear
hyaline or eosinophilic cytoplasmic
globules, resembling the appearance
of embryonic muscle cells; F, sarco-
matoid features: cell possess high-
grade spindled or polygonal mor-
phology with occasional fascicular
cellarrangementorstoriformpattern,
resembling a fibrosarcoma.
prominent
490
q 2005 Lippincott Williams & Wilkins
Upton et al
J Immunother?Volume 28, Number 5, September/October 2005

Page 4

the three prognostic groups when the best prognostic groups
were required to have ‘‘no granular features’’ versus ‘‘no or
,50% granular features.’’ The final model was selected sub-
jectively based on both heuristic considerations (eg, the fre-
quency of the potential predictor; the association of one
predictor with another; the optimal coding for the variable) and
statistical significance. Once this prognostic model was
developed,we applieditwithoutmodificationtoanindependent
sample of patients whose tumor specimens were collected from
metastatic sites as an independent assessment of its validity. In
addition, as a separate test of its validity, the model was
applied to survival data for all 231 patients. Median survivals
for the predefined predictive groups were calculated using
a log-rank test applied to the Kaplan-Meier estimated survival
distribution. P , 0.05 defined statistical significance.
RESULTS
Demographic and Specimen Features
Of the 231 specimens included in this analysis, 163
were from the kidney and 68 were from other sites. The
patient, treatment, and response characteristics for specimens
requested compared with those received and evaluable for
analysis are displayed in Table 1. Most of specimens were col-
lected from patients treated after 1996, and a disproportionate
number of patients treated at BIDMC received low-dose IL-2
regimens; otherwise there was no discernible collection bias.
Thirty-eight patients (16.5%) with specimens had responded
to IL-2-based therapy, similar to the 15.5% response rate for
the whole IL-2-treated population for which specimens were
sought. Response rates were higher for patients with specimens
who received high-dose IL-2 (23% vs. 12%) and for patients
treated who received high-dose IL-2 with unresected primary
tumors (ie, those with metastatic pathology specimens [22%
vs. 0%]), as previously reported.36
Table 2 describes the specimens analyzed, while the
sources of metastatic specimens are listed in Table 3. Twenty-
seven specimens (17 renal and 10 from metastatic sites) were
determined to be non-clear cell or indeterminate in type.
Specimens received from patients treated on BIDMC trials had
similar histologic features to those received from CWG, with
89% and 86% being clear cell, respectively.
Predicting Response from Kidney Specimens
Tumor response by various pathologic features is pre-
sented in Table 4. Tumor responses were seen in 30 of
TABLE 1. Distribution of Specimens Requested vs. Received
All Requested
Specimens
All Requested
BIDMC Specimens
All Requested
CWG Specimens
Kidney Specimens
Analyzed
Other Sites
Analyzed
No. specimens
Mean age (yr)
Gender (M/F) (% male)
IL-2 treatment
High-dose IL-2
Low-dose IL-2 6 IFN
Response rate (CR/PR)
High-dose IL-2 treatment
Low-dose IL-2 treatment
Total population
Date of treatment
1990–1996
1997–2001
444
55 6 7
56388
54 6 7
163
55 6 9
114/49 (70%)
68
59 6 10
42/14 (75%)
55 6 10
49/19 (72%) 314/130 (71%) 272/116 (70%)
206 (46%)
238 (54%)
11 (20%)
45 (80%)
195 (50%)
193 (50%)
70 (43%)
93 (57%)
32 (47%)
36 (53%)
40/206 (19%)
29/238 (12 %)
69/444 (16%)
3/11 (27%)
4/45 (8%)
7/56 (12%)
37/195 (19%)
25/193 (13%)
62/388 (16%)
16/70 (23%)
15/93 (16%)
31/163 (19%)
7/32 (22%)
0/36 (0%)
7/68 (10%)
196 (44%)
248 (56%)
0 (0%)
56 (100%)
196 (51%)
192 (49%)
14 (9%)
149 (91%)
5 (7%)
63 (93%)
TABLE 2. Description of Specimens Analyzed
Kidney SpecimensOther Sites
n
Tumor type
Clear cell
Chromophobe
Papillary type I
Papillary type II
Indeterminate
Tissue of specimen included in analysis
Kidney
Lung
Other sites
Type of specimen included in analysis
Resection
Biopsy
Fine-needle aspirate
16368
146 (90%)
2 (1%)
2 (1%)
10 (6%)
3 (2%)
58 (85%)
0
0
4 (6%)
6 (9%)
163 (100%)
16 (24%)
52 (76%)
152 (93%)
11 (7%)
3 (4%)
55 (81%)
10 (15%)0
TABLE 3. Sites of Metastatic Carcinoma Specimens
SiteNo.
Bone
Lung
Lymph node
Liver
Mucosal sites (vagina, GI tract)
Soft tissue, NOS, including ‘‘chest wall’’
Brain
Adrenal
Pleura
Skin
21 (31%)
16 (24%)
11 (16%)
6 (9%)
5 (7%)
4 (6%)
2 (3%)
1 (1%)
1 (1%)
1 (1%)
Total 68 (100%)
q 2005 Lippincott Williams & Wilkins
491
J Immunother?Volume 28, Number 5, September/October 2005 Histologic Predictors of RCC Response to IL-2

Page 5

146 patients with conventional renal carcinoma (21%)
compared with only 1 of 17 patients with non-clear cell
histology (6%) (P = 0.20). The one tumor response in a patient
with non-clear cell carcinoma occurred in a patient originally
classified as having a papillary carcinoma (chromophil
carcinoma), type II. Subsequent analysis determined that this
patient had a collecting duct carcinoma.
The arrangement of cells in clear cell carcinomas is most
commonly seen in three patterns: alveolar, tubular, and cystic.
We observed a statistically significant difference (P = 0.02) in
response rate depending on the amount of alveolar component
present in the tumors. Tumors with greater than 50% alveolar
pattern had a 28% response rate (23 of 83 cases) compared
with a 13% response rate (7 of 55 cases) in tumors with less
than 50% alveolar pattern and with no responses in five patients
with no alveolar pattern. Neither tubular morphology nor cys-
tic morphology was significantly associated with response to
IL-2-based therapy (both P . 0.25).
For clear cell carcinomas, granular features appeared to
predict a poor response to IL-2-based therapy. For tumors with
more than 50% granular morphology, the therapeutic response
rate was only 5% (1 of 19 cases) compared with 19% (13/67)
for tumors with less than 50% granular morphology and 29%
(16/56) for tumors without any granular morphology (P =
0.03). When we controlled for the presence of rhabdoid features
in our analysis, there remained an independent association
between granular features and a poor response to IL-2-based
treatment. When papillary morphology was assessed as a
morphologic component rather than as a tumor type, there was
some evidence that it was associated with a lack of response
(P = 0.20), so it was considered further when developing an
overall predictive rule. Neither the presence nor the proportion
of rhabdoid or sarcomatoid phenotypes showed any associ-
ation with response. Similarly, we did not find any association
between tumor necrosis, TILs, or nuclear grade and response
to IL-2-based therapy.
For renal specimens, we were also able to assess for
renal vein involvement. There was a high response rate among
patients with major vein only involvement, which led us to com-
pare the overall response rates of those with major vein in-
volvement(30%,18/60)tothosewithoutmajor veininvolvement
(14%, 12/86 patients, including 3 responses in 14 patients with-
out adequate material to assess this feature) (P = 0.02). As this
feature could not be applied to metastatic sites, it was not
included in subsequent models.
Application of statistical modeling techniques to these
data yielded several predictive models. The model in Table 5
was selected as it involves only four variables, provides
reasonably sized groups for each prognostic category, and has
a highly significant trend (P , 0.0001). This model classified
patients with clear cell tumors into three groups. Patients with
tumors containing papillary features or more than 50%
granular features or without alveolar features had a poor
chance of response to IL-2 (3% response rate, 1 response in
33 patients). In contrast, patients with significant (.50%) alve-
olar and no granular or papillary features had a 39% response
rate (14/36 patients). Patients with either less alveolar features
or some granular features (,50%) had an intermediate chance
of responding to IL-2 (19%, 15/77 patients). The trend across
these response groups remained highly significant (P , 0.001)
even after adjustment for type of IL-2 treatment (high vs. low
dose).
Application to Nonrenal Specimens
We then applied this model using alveolar, granular, and
papillary features to the 58 clear cell pathology specimens
obtained from metastatic sites (Table 6). Five tumor responses
were found in the 20 patients in the best prognosis group
(25%), with the other two responses occurring in the 22 pa-
tients in the intermediate prognosis group (9%). None of the
16 patients in the poor prognosis group responded (P = 0.02).
As responses occurred only among those treated with high-
dose IL-2, the analysis was repeated limited to this group, and
the trend in response rate (5/14, 36%; 2/9, 22%; and 0/9, 0% in
TABLE 4. Response Rate by Specific Factors in
Kidney Specimens
Tumor Type
Non-clear cell
Clear cell
1/17 (6%)
30/146 (21%)
Individual Factors in
Clear Cell Tumors
Coding
None
,50%
.50%
Sarcomatoid
Rhabdoid
Papillary
Granular/eosinophilic
Tubular
Alveolar
Cystic
Solid
22/104 (21%)
19/86 (22%)
30/132 (23%)
16/56 (29%)
11/61 (18%)
0/5 (0%)
26/130 (20%)
18/79 (23%)
7/23 (30%)
8/43 (19%)
0/9 (0%)
13/67 (19%)
17/61 (28%)
7/55 (13%)
4/9 (44%)
3/9 (33%)
1/15 (7%)
2/14 (14%)
0/5 (0%)
1/19 (5%)
2/20 (10%)
23/83 (28%)
0/7 (0%)
0/2 (0%)
TILS and Necrosis
Coding
NoneGrade 1 Grade 2Grade 3
TILs
Necrosis
5/16 (31%)
2/7 (29%)
6/27 (22%)
4/15 (27%)
9/32 (28%)
2/19 (11%)
2/20 (10%)
13/56 (23%)
Nuclear Fuhrman
Grade
Coding
Grade 1 Grade 2Grade 3 Grade 4
Predominant grade*
Overall grade*
2/6 (33%)10/33 (30%) 14/87 (16%)
0/0 (0%)2/10 (20%) 11/55 (20%) 17/81 (21%)
4/20 (20%)
Metastatic
Invasion
Coding
NoMissing/UnknownYes
Invasion of perinephric
tissue
Regional lymph nodes
Renal vein invasion
Renal vein invasion: by known site
No known site†
Major vein only
Minor vein only
Both major and
minor veins
7/43 (16%)
5/25 (20%)
8/48 (17%)
2/17 (12%)
16/95 (17%)
3/14 (21%)
21/86 (24%)
9/26 (35%)
19/84 (23%)
11/62 (18%)
12/25 (48%)
1/24 (4%)
6/35 (17%)
*‘‘Predominant grade’’ is the nuclear Fuhrman grade occupying the greatest
percentage area of the carcinoma. ‘‘Overall grade’’ is the highest nuclear Fuhrman grade
observed.
†Includes missing information.
492
q 2005 Lippincott Williams & Wilkins
Upton et al
J Immunother?Volume 28, Number 5, September/October 2005