- Email: [email protected]
Trends in reporting histological subtyping of renal cell carcinoma: association with cancer center type
Human Pathology (2018) 74, 99–108
www.elsevier.com/locate/humpath
Original contribution
Trends in reporting histological subtyping of renal cell carcinoma: association with cancer center type☆,☆☆ Ted Gansler MD,MBA,MPH a,⁎, Stacey Fedewa PhD, MPH a , Mahul B. Amin MD b , Chun Chieh Lin MBA, PhD a , Ahmedin Jemal DVM, PhD a a
Intramural Research, American Cancer Society,Atlanta, GA 30303 Department of Pathology and Laboratory Medicine, The University of Tennessee Health Science Center, Memphis, TN 38163
b
Received 20 October 2017; revised 3 January 2018; accepted 5 January 2018
Keywords: Renal cell carcinoma; Pathology; Classification; Registries; Healthcare quality
Summary Histological classification of renal cell carcinoma (RCC) has become increasingly important for clinical management. We identified 295 483 RCC diagnosed from 1998-2014 in the National Cancer Database (NCDB) to examine temporal trends in proportions of RCC with unspecified histology and several specific histologies from the 1998 and 2004 World Health Organization classifications of RCC. Further, multivariable log binomial analysis of 101 062 RCC diagnosed from 2010 to 2014 was used to determine whether the association of diagnosing/treating facility type and the proportion of unspecified RCC is independent of patient demographic and clinical factors. Between 1998 and 2014, the proportion of histologically unspecified RCC decreased substantially in all facility types, with the decrease smallest in community programs (from 86.0% to 28.1%) and largest in National Cancer Institute–designated centers (from 85.1% to 9.8%). These decreases were offset by increases in percentages of papillary, clear cell, and chromophobe RCC cases. During 2010 to 2014, relative to community programs, RCCs were 21% less likely to be reported as unspecified histology (adjusted prevalence ratio [aPR] = 0.79; 95% CI, 0.68-0.92) in comprehensive community programs, 32% less likely in integrated network programs (aPR = 0.68; 95% CI, 0.57-0.92) and academic programs (aPR = 0.68; 95% CI, 0.54-0.87), and 63% less likely (aPR = 0.37; 95% CI, 0.26-0.52) in National Cancer Institute –designated programs. These results have implications for the optimal selection of targeted systemic therapies for patients with advanced disease, and for the potential value of cancer registry data in pathology quality improvement programs to promote more rapid and consistent adoption of new classifications of RCC and other neoplasms. © 2018 Elsevier Inc. All rights reserved.
☆
Financial disclosures: None. This study used the National Cancer Data Base (NCDB), which is supported by the American College of Surgeons and the American Cancer Society. The authors acknowledge the efforts of the American College of Surgeons and the Commission on Cancer in the creation of the National Cancer Data Base. The data used in the study are derived from the limited data set of the NCDB. The American College of Surgeons and the Commission on Cancer have not verified and are not responsible for the analytic or statistical methodology employed, or the conclusions drawn from these data by the authors. ⁎ Corresponding author. E-mail addresses: [email protected] (T. Gansler), [email protected] (S. Fedewa), [email protected] (M. B. Amin), [email protected] (C. C. Lin), [email protected] (A. Jemal). ☆☆
https://doi.org/10.1016/j.humpath.2018.01.010 0046-8177/© 2018 Elsevier Inc. All rights reserved.
100
1. Introduction Classification of renal cell carcinoma (RCC) has become substantially more complex during the past two decades, with recognition of several new types of RCC based on progress in understanding the relationships among molecular pathology, morphology, and natural history. For more than a decade, National Comprehensive Cancer Network (NCCN) treatment guidelines have recommended that RCC histology be considered in selecting systemic molecularly targeted therapies [1], and the increasing number of drugs targeting specific molecular alterations present in subsets of RCC has further increased the importance of precise morphologic and molecular classification [2,3]. Because several RCC types are associated with inherited syndromes, precise histological characterization can also help identify families most likely to benefit from genetic counseling and testing [4-8]. In addition, some retrospective studies have noted prognostic differences among certain RCC subtypes, although these differences may not persist after consideration of other variables such as stage and grade [9-14]. Relatively little is known regarding the promptness with which the 1997 Heidelberg classification [15], the 1998 World Health Organization (WHO) classification [16], and the 2004 WHO classification [17] were adopted, and even less is known about the consistency of adopting this new terminology in various practice settings. This information could help in facilitating consistent use of new diagnostic categories introduced by the 2016 WHO classification of RCC and the 2013 International Society of Urological Pathology Vancouver classification [5-8] and, more generally, in quality measurement and quality improvement for other pathology-based disease classification systems. In a previous study using the National Cancer Database (NCDB), we observed that relative to National Cancer Institute (NCI)–designated comprehensive programs, community programs were significantly more likely to record broader diagnoses (rather than narrower diagnoses) for cancers of several organs, and this association was particularly prominent for malignant neoplasms of the kidneys [18]. In the current study, we use the NCDB records to examine trends from 1998 through 2014 in the adoption by various cancer center categories of several of the most common RCC types in the 1997 Heidelberg, 1998 WHO, and 2004 WHO classifications. We also examined whether cancer center categories are independently associated with the prevalence of reporting histologically unspecified RCC during a period (2010-2014) when consideration of histological type is recommended in the selection of targeted therapies for patients with metastatic RCC. Unlike other databases such as the NCI’s Surveillance, Epidemiology, and End Results Program, the NCDB provides information regarding the treatment center category and whether the patient was diagnosed at the treating facility [19-22].
T. Gansler et al.
2. Materials and methods 2.1. Study participants We examined records in the NCDB of patients diagnosed with RCC from 1998 through 2014. The NCDB, a collaboration of the American College of Surgeons and the American Cancer Society, collects data from more than 1 450 Commission on Cancer (CoC)–approved hospitals and approximately 70% of all cancers of the kidneys and renal pelvis diagnosed in the United States [19]. The NCDB contains data on patient demographics, clinical characteristics, and facility type that are abstracted from medical records and recorded according to a standardized data dictionary [20-22]. The Morehouse University Institutional Review Board in Atlanta, Georgia determined that this study is exempt from review. The International Classification of Diseases for Oncology version 3 (ICD-O-3) was first used by the NCDB in 2001 and although the International Classification of Diseases for Oncology version 2 (ICDO-2) included codes for papillary RCC and clear cell RCC, there was no code in the ICDO-2 for chromophobe RCC, cyst-associated RCC, sarcomatoid RCC, and renal collecting duct carcinoma [23]. By starting analyses in 1998, we were therefore able to include baseline data before adoption of ICDO-3. We limited some analyses to cases diagnosed from 2010 through 2014 to reflect patterns of pathology interpretation during the current era in which RCC histologic types are recognized by oncology guidelines as a factor in choosing systemic therapy for patients with metastatic RCC [1]. We initially selected patients diagnosed and reported by the same CoC-accredited facility, with a malignant neoplasm of the kidney confirmed by histology, and with ICD-O-3 histology codes 8260 (papillary adenocarcinoma, NOS), 8310 (clear cell adenocarcinoma, NOS), 8312 (renal cell carcinoma), 8316 (cyst-associated renal cell carcinoma), 8317 (renal cell carcinoma, chromophobe type), 8318 (renal cell carcinoma, sarcomatoid), 8319 (collecting duct carcinoma), or 8320 (granular cell carcinoma). Although there are several other types of RCC in the 2004 WHO classification, such as renal carcinomas associated with Xp11.2 translocation/TFE3 gene fusions, RCC associated with neuroblastoma, and mucinous tubular and spindle cell carcinoma, there are no specific ICDO-3 codes to represent these histologies; therefore, these cases could not be identified in NCDB records. Thus, the ICD-O-3 histology code, 8312, includes (1) RCC cases diagnosed only as “RCC” by the pathologist but which actually had specific histologies with corresponding ICDO-3 codes (such as papillary, clear cell, and chromophobe), (2) a much smaller number of cases with rare RCC histologies that may have been diagnosed by the pathologist but which were assigned a more general code by the registry because there are no specific ICDO-3 codes for these entities, and (3) a small number of cancers meeting criteria for unclassified RCC, a diagnosis for which there is no corresponding ICDO-3 code.
RCC histology trends Table 1
101
Exclusion criteria for renal cell carcinoma cases, National Cancer Database 1998-2014 and 2010-2014
Included and excluded cases
No. 1998-2014
No. 2010-2014
Initial cases of biopsy-confirmed renal cell carcinoma, diagnosed and reported at same CoC-accredited facility Exclude missing sex Exclude missing race Exclude age b20 and age ≥90 Exclude missing AJCC stage group stage or stage 0 Exclude missing summary surgery of primary site or no nephrectomy Exclude “other” or pediatric facility type Final
385542
130093
95 25788 442 + 1132 26107 + 3 134 + 45731 972 295483 (76.6%)
22 4894 146 + 409 7687 + 1 39 + 18473 200 101062 (77.7%)
NOTE. Some cases had more than one exclusion criterion, so subtracting individual exclusions from the initial cases does not exactly equal the final sample. Abbreviations: AJCC, American Joint Committee on Cancer; CoC, Commission on Cancer.
2.2. Outcome and independent variables The primary outcome was histologic type of RCC. Because most histologic types of RCC are too rare for sufficiently precise analyses even in this large data set, we focused on histologically unspecified RCC (coded as RCC without any specific type), and the 3 most common specific types—papillary, clear cell, and chromophobe RCC. Although sarcomatoid RCCs were included among the histologies for including RCC cases, the Table 2
diagnosis of sarcomatoid RCC was not an outcome in our analyses because the Heidelberg classification and the 2004 WHO classification do not consider it as a specific type of RCC, but instead as a secondary change that can occur in RCC with various histologic types (for example, clear cell RCC with focal sarcomatoid change). Similarly, granular RCC was included in the denominator for RCC cases, but not individually reported as an outcome because it is considered an obsolete diagnosis and is not included in the Heidelberg classification and the
Patient characteristics by facility type among renal cell carcinoma cases, National Cancer Database 1998-2014 CP, n (%)
Year group (P b .001) 1998-2000 3653 (14.8) 2001-2005 7686 (31.2) 2006-2010 7616 (30.9) 2011-2014 5674 (23.0) Age group (P b .001) 20-39 years 1005 (4.1) 40-49 years 2985 (12.1) 50-59 years 5873 (23.8) 60-69 years 7120 (28.9) 70-79 years 5715 (23.2) 80-89 years 1932 (7.8) Sex (P b .001) Male 14965 (60.8) Female 9665 (39.2) Race/ethnicity (P b .001) Non-Hispanic white 20337 (82.6) Hispanic 1369 (5.6) Black 2249 (9.1) Asian and PI 406 (1.6) Other 269 (1.1) AJCC stage group (P b .001) I 15690 (63.7) II 3359 (13.6) III 3862 (15.7) IV 1719 (7.0) Total 24630
CCP, n (%)
INP, n (%)
ACADP, n (%)
NCIP, n (%)
All Facilities, n (%)
15400 (12.4) 35319 (28.4) 40538 (32.6) 33030 (26.6)
4418 (12.4) 10420 (29.3) 11671 (32.9) 9016 (25.4)
8292 (11.1) 19492 (26.1) 23936 (32.1) 22955 (30.7)
3648 (10.0) 9703 (26.7) 11504 (31.6) 11511 (31.7)
35411 (12.0) 82620 (28.0) 96266 (32.2) 82186 (27.8)
5296 (4.3) 15209 (12.2) 30465 (24.5) 36167 (29.1) 28473 (22.9) 8677 (7.0)
1694 (4.8) 4811 (13.5) 9131 (25.7) 10295 (29.0) 7525 (21.2) 2069 (5.8)
4322 (5.8) 11026 (14.8) 20072 (26.9) 21157 (28.3) 14323 (19.2) 3775 (5.1)
2189 (6.0) 5354 (14.7) 9670 (26.6) 10924 (30.0) 6756 (18.6) 1473 (4.1)
14506 (4.9) 39385 (13.3) 75211 (25.5) 85663 (29.0) 62792 (21.3) 17926 (6.1)
76475 (61.5) 47812 (38.5)
21981 (61.9) 13544 (38.1)
46247 (61.9) 28428 (38.1)
23638 (65.0) 12728 (35.0)
183306 (62.0) 112177 (38.0)
103170 (83.0) 6676 (5.4) 11589 (9.3) 1617 (1.3) 1235 (1.0)
26929 (75.8) 2542 (7.2) 5252 (14.8) 428 (1.2) 374 (1.1)
54424 (72.9) 5362 (7.2) 12127 (16.2) 1621 (2.2) 1141 (1.5)
28876 (79.4) 1630 (4.5) 4591 (12.6) 790 (2.2) 479 (1.3)
233736 (79.1) 17579 (5.9) 35808 (12.1) 4862 (1.6) 3498 (1.2)
82188 (66.1) 16199 (13.0) 18486 (14.9) 7414 (6.0) 124287
23730 (66.8) 4348 (12.1) 5409 (15.2) 2038 (5.7) 35525
50533 (67.7) 8371 (11.2) 11248 (15.5) 4523 (6.1) 74675
25079 (69.0) 3367 (9.3) 5653 (15.5) 2267 (6.2) 36366
197220 (66.7) 35644 (12.1) 44658 (15.1) 17961 (6.1) 295483
NOTE. The χ2 tests and the corresponding two-tailed P-values represent differences in facility type by patient characteristics. Abbreviations: CP, Community Cancer Program; CCP, Comprehensive Community Cancer Program; INP, Integrated Network Cancer Program; ACADP, Academic Cancer Program; NCIP, National Cancer Institute-Designated Cancer Program; PI, Pacific Islander.
102
T. Gansler et al.
2004 WHO classification [15-17]. Collecting duct carcinoma diagnoses were not studied because there were too few cases for precise analysis. For some analyses, we dichotomized the histology variable to unspecified RCC versus any specific RCC histology (excluding, as noted above, sarcomatoid and granular RCC). Our primary independent variable of interest was facility type. Five facility types were studied—community programs, comprehensive community programs, integrated network programs, academic programs, and NCI-designated programs. Criteria for cancer center classification are described in detail elsewhere [24,25]. Patients reported by Pediatric programs and “other” program types were excluded because of small numbers. Briefly, the spectrum from community programs to comprehensive community programs and integrated network programs, to academic programs, and to NCI-designated programs generally reflects greater comprehensiveness of diagnostic and treatment services, numbers of patients treated, participation in clinical research, and involvement in postgraduate training of physicians. Our interest in this variable is based largely on our prior study of cases diagnosed prior to widespread use of targeted therapies for metastatic RCC, showing a higher percentage of unspecified RCC cases reported by community programs and comprehensive community programs Table 3
relative to academic programs and NCI-designated programs [18]. This interest is supported by prior reports of differences in RCC histology interpretation between records from cancer registries and review of the same specimens by subspecialists in renal or urological pathology [26]. Also, the percentage of cases recorded only as “RCC” or as “RCC not otherwise specified” is much higher in reports based on cancer registries (which include cases from all facility types) than in series of cases from individual academic cancer centers or groups of several academic centers [10,14,27-29]. We also examined variables that might potentially be associated with histology and facility type, such as year of diagnosis, patient age and sex, patient race/ethnicity, and American Joint Committee on Cancer (AJCC) stage group; cases missing values for these variables were therefore excluded as shown in Table 1. The stage group recorded in the NCDB for each patient was based on the AJCC version that was in effect at the time of diagnosis; the 7th edition was used 20102014, the 6th edition staging was used in 2003-2009, and 5th edition staging was used in 1998-2002. Patients younger than 20 years and those older than 89 years were excluded because of small numbers in these groups. Finally, patients who did not undergo nephrectomy (partial, total, radical, or unspecified)
Patient characteristics by facility type among renal cell carcinoma cases, National Cancer Database 2010 to 2014 CP, n (%)
Year (P b .001) 2010 1446 (20.3) 2011 1390 (19.5) 2012 1449 (20.4) 2013 1367 (19.2) 2014 1468 (20.6) Age group (P b .001) 20-39 years 315 (4.4) 40-49 years 863 (12.1) 50-59 years 1768 (24.8) 60-69 years 2179 (30.6) 70-79 years 1530 (21.5) 80-89 years 465 (6.5) Sex (P b .001 Male 4428 (62.2) Female 2692 (37.8) Race/ethnicity (P b .001) Non-Hispanic white 5631 (79.1) Hispanic 509 (7.1) Black 747 (10.5) Asian & PI 138 (1.9) Other 95 (1.3) AJCC stage group (P b .001) I 4775 (67.1) II 900 (12.6) III 1041 (14.6) IV 404 (5.7) Total 7120
CCP, n (%)
INP, n (%)
ACADP, n (%)
NCIP, n (%)
All Facilities, n (%)
7820 (19.1) 7952 (19.5) 8195 (20.1) 8477 (20.8) 8406 (20.6)
2167 (19.4) 2313 (20.7) 2279 (20.4) 2188 (19.6) 2236 (20.0)
4975 (17.8) 5549 (19.9) 5742 (20.6) 5822 (20.8) 5842 (20.9)
2468 (17.7) 2601 (18.6) 2987 (21.4) 2931 (21.0) 2992 (21.4)
18876 (18.7) 19805 (19.6) 20652 (20.4) 20785 (20.6) 20944 (20.7)
1897 (4.6) 4834 (11.8) 10298 (25.2) 12794 (31.3) 8537 (20.9) 2490 (6.1)
567 (5.1) 1482 (13.3) 2909 (26.0) 3459 (30.9) 2238 (20.0) 528 (4.7)
1686 (6.0) 4079 (14.6) 7626 (27.3) 8442 (30.2) 4896 (17.5) 1201 (4.3)
875 (6.3) 1965 (14.1) 3831 (27.4) 4434 (31.7) 2394 (17.1) 480 (3.4)
5340 (5.3) 13223 (13.1) 26432 (26.2) 31308 (31.0) 19595 (19.4) 5164 (5.1)
25345 (62.0) 15505 (38.0)
7054 (63.1) 4129 (36.9)
17448 (62.5) 10482 (37.5)
9130 (65.3) 4849 (34.7)
63405 (62.7) 37657 (37.3)
33052 (80.9) 2636 (6.5) 4044 (9.9) 610 (1.5) 508 (1.2)
8354 (74.7) 803 (7.2) 1744 (15.6) 152 (1.4) 130 (1.2)
19364 (69.3) 2353 (8.4) 4888 (17.5) 765 (2.7) 560 (2.0)
10864 (77.7) 734 (5.3) 1856 (13.3) 311 (2.2) 214 (1.5)
77265 (76.5) 7035 (7.0) 13279 (13.1) 1976 (2.0) 1507 (1.5)
28595 (70.0) 4520 (11.1) 5779 (14.1) 1956 (4.8) 40850
7834 (70.1) 1189 (10.6) 1670 (14.9) 490 (4.4) 11183
19749 (70.7) 2575 (9.2) 4280 (15.3) 1326 (4.7) 27930
9973 (71.3) 1098 (7.9) 2243 (16.0) 665 (4.8) 13979
70926 (70.2) 10282 (10.2) 15013 (14.9) 4841 (4.8) 101062
NOTE. The χ2 tests and the corresponding 2-tailed P-values represent differences in facility type by patient characteristics. Abbreviations: CP, Community Cancer Program; CCP, Comprehensive Community Cancer Program; INP, Integrated Network Cancer Program; ACADP, Academic Cancer Program; NCIP, National Cancer Institute–Designated Cancer Program; PI, Pacific Islander.
RCC histology trends
103
were excluded because diagnosis of these cases as unspecified RCC might have been influenced by the amount of tissue available (especially with core biopsies), and our intent was to examine factors entirely under the pathologist’s control.
2.3. Data analysis We performed bivariate statistical analyses with SPSS Statistics version 24 (Armonk, NY). χ2 Tests and the corresponding two-tailed P-values with a significance level at .05 were used to examine potential associations of RCC histology with the independent variables noted above. Multivariable logbinomial analyses were performed with SAS, version 9.3 (SAS, Cary, NC). Adjusted multivariable log binomial models, which accounted for clustering at the facility level (correlation of characteristics among patients treated at a particular hospital) were used to estimate adjusted prevalence ratios (aPR) and 95% confidence intervals (CI) for unspecified RCC versus all other RCC types combined [30].
3. Results 3.1. Study participants and characteristics, 1998-2014 Of the initial 385,542 cases of RCC recorded in the NCDB from 1998 through 2014, 295,483 (76.6%) remained in the
analytic sample after exclusions; whereas 101 062 (77.7%) of the 130 093 cases diagnosed from 2010 through 2014 remained after exclusions (Table 1). Characteristics of the analytic sample for RCC cases diagnosed from 1998 through 2014 and for the subset diagnosed from 2010 through 2014 are summarized in Tables 2 and 3, respectively. During both time periods, the modal age category was 60 to 69 years, a majority of patients were male, and more than three fourths were non-Hispanic whites. Approximately two-thirds of patients had AJCC stage I disease. There were statistically significant associations between facility type and all-patient sociodemographic and clinical characteristics, but the magnitudes of these differences generally were small except for patient age.
3.2. Trends in percentages of RCC with specific histologies and unspecified histology, 1998-2014 The percentage of histologically unspecified RCC steadily decreased between 1998 and 2014, with corresponding increases in percentages of papillary, clear cell, and chromophobe RCC cases (Figure). Unspecified RCC comprised 86.0% of all RCC diagnosed by community programs in 1998, but only 28.1% in 2014. This decline was even more prominent in NCI-designated programs (85.1% in 1998 versus 9.8% in 2014), and was intermediate for other facility categories (for example, 86.8% in 1998 versus 15.4 in 2014 in integrated network programs).
25
70 60
% Clear Cell RCC
% Papillary RCC
20
15
10
50
40 30 20
5 10 0 1998
2000 CP
2002
2004
CCP
2006 INP
2008
2010
ACADP
2012
0 1998
2014
% RCC with Unspecified Histology
9
% Chromophobe RCC
2002
2004
CCP
2006 INP
2008
2010
ACADP
2012
2014
NCIP
100
10
8 7 6 5 4 3 2 1 0 1998
2000 CP
NCIP
2000 CP
2002 CCP
2004
2006 INP
2008
2010
ACADP
2012 NCIP
2014
90 80 70 60 50 40 30 20 10 0 1998
2000 CP
2002 CCP
2004
2006 INP
2008
2010
ACADP
2012
2014
NCIP
Figure Time trends in percentage of renal cell carcinoma cases by histologic type. Abbreviations: ACADP, Academic (teaching/research) Cancer Programs; CCP, Comprehensive Community Cancer Programs; CP, Community Cancer Programs; INP, Integrated Network Cancer Programs; NCIP, National Cancer Institute-designated Cancer Programs, RCC, renal cell carcinoma.
104
T. Gansler et al.
For all years from 2002 through 2014, NCI-designated programs diagnosed the smallest percentage of RCC cases with unspecified histology, and the largest percentages of papillary, clear cell, and chromophobe RCC cases. Conversely, community programs diagnosed the largest percentage of RCC cases with unspecified histology and the smallest percentage of chromophobe RCC and papillary RCC for all years during this interval (albeit, tied with comprehensive community programs during 2012 for the lowest percentage of papillary RCC), as well as the smallest percentage of clear cell RCC during 10 of these 13 years. During all years from 2002 through 2014, academic programs had the second or third lowest percentages of RCC
Table 4
with unspecified histology. Academic centers had the second or third highest percentages during each of these years for papillary and chromophobe RCC, and during 10 of these 13 years for clear cell RCC. A graph for granular RCC is not shown in the Figure because this term is not used in the 2004 WHO classification. As expected, however, we observed a steady decline in use of this term, from 1.8% of all RCC diagnoses during 1998 to 2000, to 0.1% during 2011 through 2014. Overall, this diagnosis was used least often in NCI-designated programs during this period (0.4%) and most often (0.8%) by community programs, comprehensive community programs, and integrated network programs (not shown).
Patient, tumor, and facility characteristics by renal cell carcinoma histology, National Cancer Database 2010 to 2014 Papillary N (%)
Year (P b.001) 2010 2684 (14.2) 2001 2921 (14.7) 2012 3183 (15.4) 2013 3183 (15.3) 2014 3170 (15.1) Age group (P b.001) 20-39 years 519 (9.7) 40-49 years 1609 (12.2) 50-59 years 3898 (14.7) 60-69 years 5131 (16.4) 70-79 years 3208 (16.4) 80-89 years 776 (15.0) Sex (P b.001) Male 11362 (17.9) Female 3779 (10.0) Race/ethnicity (P b.001) Non-Hispanic white 9880 (12.8) Hispanic 451 (6.4) Black 4500 (33.9) Asian & PI 182 (9.2) Other 128 (8.5) AJCC stage group (P b.001) I 11902 (16.8) II 1596 (15.5) III 1341 (8.9) IV 302 (6.2) Facility type (P b.001) CP 841 (11.8) CCP 5216 (12.8) INP 1805 (16.1) ACADP 4658 (16.7) NCIP 2621 (18.7) Total 15141 (15.0)
Clear cell (%)
Chromophobe N (%)
Any specific type N (%)
Unspecified N (%)
10331 (54.7) 11128 (56.2) 11679 (56.6) 12099 (58.2) 12399 (59.2)
1079 (5.7) 1177 (5.9) 1274 (6.2) 1310 (6.3) 1309 (6.3)
14260 (75.6) 15409 (77.8) 16288 (78.9) 16753 (80.6) 17022 (81.3)
4348 (23.0) 4167 (21.0) 4143 (20.1) 3799 (18.3) 3711 (17.7)
3048 (57.1) 7729 (58.5) 15250 (57.7) 17731 (56.6) 10984 (56.1) 2894 (56.0)
537 (10.1) 946 (7.2) 1521 (5.8) 1639 (5.2) 1179 (6.0) 327 (6.3)
4167 (78.0) 10429 (78.9) 20876 (79.0) 24732 (79.0) 15505 (79.1) 4023 (77.9)
1138 (21.3) 2664 (20.1) 5239 (19.8) 6213 (19.8) 3848 (19.6) 1066 (20.6)
35045 (55.3) 22591 (60.0)
3452 (5.4) 2697 (7.2)
50340 (79.4) 29392 (78.1)
12285 (19.4) 7883 (20.9)
46093 (59.7) 4702 (66.8) 4688 (35.3) 1270 (64.3) 883 (58.6)
4627 (6.0) 418 (5.9) 862 (6.5) 137 (6.9) 105 (7.0)
61138 (79.12) 5614 (79.8) 10258 (77.3) 1599 (80.9) 1123 (74.5)
15248 (19.7) 1344 (19.1) 2863 (21.6) 353 (17.9) 360 (23.9)
40184 (56.7) 5351 (52.0) 9256 (61.7) 2845 (58.5)
4256 (6.0) 1055 (10.3) 770 (5.1) 68 (1.4)
56948 (80.3) 8062 (78.4) 11450 (76.3) 3273 (67.6)
13715 (19.3) 2106 (20.5) 3212 (21.4) 1135 (23.4)
3782 (53.1) 23585 (57.7) 6243 (55.8) 15507 (55.5) 8519 (60.9) 57636 (57.0)
321 (4.5) 2218 (5.4) 725 (6.5) 1784 (6.4) 1101 (7.9) 6149 (6.1)
4982 (70.0) 31300 (76.6) 8875 (79.4) 22174 (79.4) 12401 (88.7) 79732 (78.9)
2024 (28.4) 9087 (22.2) 2168 (19.4) 5439 (19.5) 1450 (10.4) 20168 (20.0)
NOTE. The χ2 tests and the corresponding 2-tailed P values represent differences between any specific histological type and unspecified RCC by patient characteristics. The “any specific type” category includes papillary, clear cell, and chromophobe RCC as well as collecting duct carcinoma, which is not shown separately because of too few cases. The total of “any specific type” and “unspecified” is slightly less than 100% because neither group includes granular RCC or sarcomatoid RCC. Please refer to the Methods section for a complete explanation. Abbreviations: AJCC, American Joint Committee on Cancer; CP, Community Cancer Program; CCP, Comprehensive Community Cancer Program; INP, Integrated Network Cancer Program; ACADP, Academic Cancer Program; NCIP, National Cancer Institute–Designated Cancer Program; PI, Pacific Islander.
RCC histology trends
3.3. Bivariate associations with RCC histology, 2010-2014 From 2010 through 2014, 79 732 cases (78.9%) were reported as a specific RCC type (excluding, as noted previously, 1162 cases with granular or sarcomatoid histologies), whereas 20168 (20.0%) were reported as histologically unspecified RCC (Table 4). Aided by this very large sample size, all independent variables were very significantly (P b .001) associated with differences in proportions of RCC with any specific histology versus unspecified RCC. However, the magnitudes of these associations seemed clinically significant for only a few of these variables. The percentages of unspecified RCC cases decreased from 23.0% in 2010 to 17.7% in 2014, with corresponding increases in the percentages of papillary, clear cell, and chromophobe RCC cases. Relative to males, females were much less often diagnosed with papillary RCC and more likely to have clear cell or chromophobe RCC. Blacks were by far the most likely (33.3%) to be diagnosed with papillary RCC, whereas this histology was least common among Hispanics (6.4%). Conversely, clear cell RCC was least common among Blacks (35.3%) and most common among Hispanics (66.8%). Stage IV RCC cases were least likely (67.6%) to be assigned a specific histology, and stage I RCCs were the most likely (80.3%), primarily due to different percentages of papillary RCC. RCC cases diagnosed in NCI-designated programs were the least likely (10.4%) and those from community programs the most likely (28.4%) to have unspecified histology. In a supplemental analysis (full results not shown) of RCC cases that were diagnosed and classified based on a biopsy specimen and which had not undergone any surgical treatment (but which otherwise met the same inclusion criteria of our main analysis), the percentages of histologically unspecified RCC cases from various facility types showed a similar pattern, with the main difference being consistently larger proportion of unspecified RCC cases. Biopsy-only cases from community programs were the most likely (57.0% of 1042) to have unspecified histology; those from NCIdesignated programs were the least likely (32.0% of 993); other facility categories had intermediate values (comprehensive community programs, 51.7% of 4079; integrated network programs, 49.0% of 940; and academic programs, 50.1%of 2209).
3.4. Independent associations with percentages of unspecified RCC, 2010-2014 A log-binomial model, accounting for clustering at the hospital level and including potential confounding covariates such as year of diagnosis, age, sex, race/ethnicity, and AJCC stage group demonstrated an independent, statistically significant, and clinically significant association between facility type and the proportion of RCC cases with unspecified histology (Table 5). Relative to community programs, the prevalence of a diagnosis of unspecified RCC in other facility types was significantly lower, with aPRs of 0.79 (95% CI, 0.68-0.92)
105 Table 5 Adjusted prevalence ratio of histologically unspecified renal cell carcinoma, National Cancer Database 2010-2014
Year 2010 2011 2012 2013 2014 Age group 20-39 years 40-49 years 50-59 years 60-69 years 70-79 years 80-89 years Sex Male Female Race/Ethnicity Non-Hispanic white Hispanic Black Asian & PI Other AJCC stage group I II III IV Facility type CP CCP INP ACADP NCIP
Prevalence ratio
95% Confidence interval
P
1.28 1.18 1.14 1.03 referent
1.17-1.41 1.09-1.28 1.05-1.23 0.97-1.09
b.001 b.001 b.001 .362
1.12 1.05 1.02 1.01 0.99 referent
1.02-1.23 0.97-1.13 0.95-1.09 0.95-1.08 0.93-1.06
.014 .255 .610 .663 .760
Referent 1.07
1.04-1.10
b.001
referent 0.94 1.12 0.93 1.23
0.82-1.08 1.04-1.21 0.74-1.18 1.07-1.40
.40 .002 .57 .003
0.77-0.88 0.79-0.91 0.87-1.00
b.001 b.001 .05
0.68-0.92 0.57-0.92 0.54-0.87 0.26-0.52
.003 .002 b.001 b.001
0.82 0.85 0.93 referent referent 0.79 0.68 0.68 0.37
NOTE. Results are based on a multivariable log-binomial model. Abbreviations: AJCC, American Joint Committee on Cancer; CP, Community Cancer Program; CCP, Comprehensive Community Cancer Program; INP, Integrated Network Program; ACADP, Academic Cancer Program; NCIP, National Cancer Institute–Designated Cancer Program; PI, Pacific Islander.
for comprehensive community programs, 0.68 for integrated network programs (95% CI, 0.57-0.92), 0.68 for academic programs (95% CI, 0.54-0.87), and most notably, 0.37 (95% CI, 0.26-0.52) for NCI-designated programs. Although most values for all covariates were significantly different from the referent categories, their effect sizes were much smaller than those for facility type.
4. Discussion This study of NCDB records demonstrated a dramatic decline between 1998 and 2014 in the percentage of RCC cases
106 recorded with unspecified histology, accompanied by inverse trends in the percentages of papillary, clear cell, and chromophobe RCC cases. These trends were most prominent and occurred earliest at NCI-designated cancer programs and latest at community programs. An adjusted analysis of RCC cases diagnosed from 2010 through 2014 demonstrated that facility type was strongly associated with the likelihood of RCC being diagnosed as having unspecified histology, independent of patient demographic and clinical factors. Despite recommendations from the 1997 Heidelberg [15], the 1998 WHO [16] and the 2004 WHO classifications [17] that specific histological types including clear cell, papillary, and chromophobe RCC be reported as such rather than as unspecified RCC, adoption occurred slowly and inconsistently. One likely reason for the persistently high percentage of histologically unspecified cases during this period is that until recently, although specific histological diagnoses had implications for prognosis and for genetic counseling of some patients [4-14], they did not influence treatment decisions. However, beginning approximately in 2007 [1], recommendations for molecularly targeted therapies for metastatic RCC varied among some specific RCC types, so it is somewhat surprising that 20% of RCC cases were unspecified during the period from 2010 through 2014. This percentage seems even more surprising given that most clear cell, papillary, and chromophobe RCCs can be recognized on routine hematoxylin and eosin–stained sections, and that in cases with equivocal histology, subtyping can usually be accomplished by the use of readily available techniques such as immunohistochemistry and fluorescence in situ hybridization [8,31,32]. Although the NCDB does not contain information on each center’s access to more sophisticated molecular testing or use of such testing for individual cases, our interpretation of these results is that the differences noted above primarily reflect variations in pathologists’ interpretation and classification of specimens processed using widely available technology. The 2016 WHO classification notes that unclassified RCC is a cancer with morphological, immunohistochemical, and chromosomal/molecular characteristics that do not correspond to any of the known common, rare, and recently described types of RCC, or to other cancers in the kidneys, such as urothelial carcinoma or metastatic cancer. When this definition is applied, only 5% to 7% of RCC should remain unclassified, a proportion much smaller than the 20% of histologically unspecified RCC in this study from 2010 through 2014 [33]. Based on the substantial difference between these two percentages, we conclude that the majority of histologically unspecified RCC do not meet WHO criteria for diagnosis as unclassified RCC. The gradual and inconsistent adoption of the 1997 Heidelberg classification and the 1998 and 2004 WHO classifications seems consistent with the literature on the “diffusion of innovation” in healthcare [34]. The consensus of published evidence, albeit largely limited to adoption of therapeutic interventions rather than diagnostic classifications, suggests that earlier adoption of medical innovations occurs in academic/research/referral institutions in contrast to community
T. Gansler et al. settings, and that complete and consistent adoption often requires several years. This pattern is consistent with our observation that the transition from “lumping” to “splitting” in RCC diagnosis occurred several years earlier and occurred most completely in NCI-designated programs relative to community programs, with an intermediate pattern among other facility types. The observational design of this study and its use of secondary data from the NCDB raise some concern regarding confounding by differences in patient demographic and clinical characteristics that might be associated with histology. These differences might arise because of differences in populations served or from selective referral of certain histological types. However, we addressed concerns regarding referral bias by (1) restricting all analyses to cases that were diagnosed and received their initial course of treatment at the same center, thereby excluding patients who were referred based on initial RCC histology and had histology re-classified by the second facility and (2) demonstrating that differences in percentages of histologically unspecified RCC among various facility types are independent of year of diagnosis; patient age, sex, and race; and AJCC stage group. Furthermore, one might expect advanced, aggressive, poorly differentiated RCC to be referred to larger and more specialized centers, but if this referral pattern occurred, the effect would be a larger rather than smaller proportion of RCC cases with unspecified histology. In adjusted models including these covariates, unspecified RCC is diagnosed approximately one third as often in NCIdesignated programs as in community programs (aPR = 0.37). In addition, relative to diagnoses in community programs, the likelihood of an unspecified RCC diagnosis is significantly lower in all other facility types, and approximately one fourth to one-third less common among comprehensive community programs (PR = 0.79), integrated network programs (aPR = 0.68), and academic programs (aPR = 0.68). Furthermore, patients with cancers of some other sites may require treatment that depends on specialized radiotherapy instruments or surgical expertise that is specific to that site and type (and therefore can lead to referral bias that limits analyses of classification according to facility types), but this is not the case with RCC. Estimating the impact, if any, of these diagnostic variations on health outcomes such as survival is beyond the scope of this investigation. However, it seems very possible that inappropriate “lumping” of some cases as unspecified RCC could have presented challenges to appropriate use of molecularly targeted systemic therapies and in identifying families that might benefit from genetic counseling relevant to hereditary cancer syndromes. It is possible that some of the variation in diagnostic categories among facilities and facility types is due to differences in abstraction of pathology reports by facility cancer registries. However, registry quality control procedures that are mandatory for CoC-approved facilities that submit cases to the NCDB most likely limit this effect, if any, to a small fraction of the observed variation. It is also possible that some communication regarding specific RCC
RCC histology trends histologic types might occur between pathologists and the oncologists who choose targeted therapies outside of formal pathology reports (ie, during multidisciplinary case management conferences, and during conversations in person or by telephone). Despite these limitations, we conclude that the results of these analyses are sufficiently robust to justify consideration of use of similar analyses in quality of cancer care measurement and quality improvement programs. However, an important limitation regarding use of registry data for monitoring the adoption of new disease classification systems is that the coding schema used by registries must include codes specific to the new diagnostic categories. Although the IDCO-3 classification used by registries included most of the histological categories in the 2004 WHO classification, it does not include most of the categories introduced by the WHO in 2016. The practical interpretation of these results is, as we have emphasized in prior studies of other cancers, not to denigrate the value of or challenges faced by pathologists in community practice settings [18,35]. Rather, our goal is to demonstrate the value of cancer registry data in identifying variations in practice patterns that might be efficiently and effectively minimized by a combination of feedback to individual centers comparing their data to benchmark values and availability of relevant continuing education programs. In fact, one of our observations in this study of RCC classification (data not shown) and in our previous studies [18,35] of other aspects of oncological pathology practice is that although there are highly significant differences among facility types, there is also prominent variation within each facility category. Thus, these data are relevant to improving consistency of histologic diagnoses for all facility categories. This strategy for quality measurement applies to a limited scope of practice pattern in surgical pathology and is expected to augment rather than replace ongoing programs of pathology organizations. However, as new technologies enhance transfer of data from medical records and laboratory information systems, this strategy might be extended to additional aspects of anatomical pathology practice.
5. Conclusions This study demonstrates substantial variations among cancer center categories in the promptness and completeness with which they utilized several diagnostic entities introduced by the 1997 Heidelberg and the 1998 WHO classifications of RCC. This diagnostic variation persists in the most recent data available (2014) and may limit application of oncology guidelines that recommend consideration of RCC histological type in selecting molecularly targeted drugs. More broadly, these results demonstrate the potential value of cancer registry data (contingent on registry use of a coding schema that reflects current diagnostic terminology) in some aspects of oncological pathology quality measurement.
107
References [1] Motzer TJ, Bolger GB, Boston B, et al. Kidney Cancer Version 2.2007. NCCN Clinical Practice Guidelines in Oncology. nccn.org; 2007. Accessed 2007. [2] Choueiri TK, Motzer RJ. Systemic therapy for metastatic renal-cell carcinoma. N Engl J Med 2017;376:354-66. [3] Motzer RJ, Jonasch E, Agarwal N, et al. Kidney Cancer Version 2.2017. NCCN Clinical Practice Guidelines in Oncology 2017. nccn.org, Accessed date: 17 April 2017. [4] Leung C, Pan S, Shuch B. Management of renal cell carcinoma in young patients and patients with hereditary syndromes. Curr Opin Urol 2016; 26:396-404. [5] Srigley JR, Delahunt B, Eble JN, et al. The International Society of Urological Pathology (ISUP) Vancouver Classification of Renal Neoplasia. Am J Surg Pathol 2013;37:1469-89. [6] Udager AM, Mehra R. Morphologic, molecular, and taxonomic evolution of renal cell carcinoma: a conceptual perspective with emphasis on updates to the 2016 World Health Organization Classification. Arch Pathol Lab Med 2016;140:1026-37. [7] Moch H, Cubilla AL, Humphrey PA, Reuter VE, Ulbright TM. The 2016 WHO Classification of Tumours of the Urinary System and Male Genital Organs—Part A: Renal, Penile, and Testicular Tumours. Eur Urol 2016; 70:93-105. [8] Moch H, Amin MB, Argani P, et al. Renal cell tumours introduction. In: Moch H, Humphrey PA, Ulbright TM, Reuter VE, editors. World Health Organization Classification of Tumours. Pathology and Genetics of Tumours of the Urinary System and Male Genital Organs. Lyon, France: IARC Press; 2016. p. 14-7. [9] Moch H, Gasser T, Amin MB, Torhorst J, Sauter G, Mihatsch MJ. Prognostic utility of the recently recommended histologic classification and revised TNM staging system of renal cell carcinoma: a Swiss experience with 588 tumors. Cancer 2000;89:604-14. [10] Amin MB, Amin MB, Tamboli P, et al. Prognostic impact of histologic subtyping of adult renal epithelial neoplasms: an experience of 405 cases. Am J Surg Pathol 2002;26:281-91. [11] Nese N, Paner GP, Mallin K, Ritchey J, Stewart A, Amin MB. Renal cell carcinoma: assessment of key pathologic prognostic parameters and patient characteristics in 47,909 cases using the National Cancer Data Base. Ann Diagn Pathol 2009;13:1-8. [12] Keegan KA, Schupp CW, Chamie K, Hellenthal NJ, Evans CP, Koppie TM. Histopathology of surgically treated renal cell carcinoma: survival differences by subtype and stage. J Urol 2012;188:391-7. [13] Tilki D, Nguyen HG, Dall'Era MA, et al. Impact of histologic subtype on cancer-specific survival in patients with renal cell carcinoma and tumor thrombus. Eur Urol 2014;66:577-83. [14] Patard JJ, Leray E, Rioux-Leclercq N, et al. Prognostic value of histologic subtypes in renal cell carcinoma: a multicenter experience. J Clin Oncol 2005;23:2763-71. [15] Kovacs G, Akhtar M, Beckwith BJ, et al. The Heidelberg classification of renal cell tumours. J Pathol 1997;183:131-3. [16] Storkel S, Eble JN, Adlakha K, et al. Classification of renal cell carcinoma: Workgroup No. 1. Union Internationale Contre le Cancer (UICC) and the American Joint Committee on Cancer (AJCC). Cancer 1997;80:987-9. [17] Eble JN, Sauter G, Epstein JI, Sesterhenn IA, editors. Pathology and genetics of tumours of the urinary system and male genital organs. World Health Organization Classification of Tumours. Lyon, France: IARC Press; 2004. p. 10. [18] Gansler T, Fedewa SA, Flanders D, Virgo KS, Ward EM. “Lumping” vs “splitting” in oncologic pathology: association with cancer center type and case volume. J Registry Manag 2012;39:43-52. [19] Lerro CC, Robbins AS, Phillips JL, Stewart AK. Comparison of cases captured in the National Cancer Data Base with those in populationbased central cancer registries. Ann Surg Oncol 2013;20:1759-65. [20] Boffa DJ, Rosen JE, Mallin K, et al. Using the National Cancer Database for Outcomes Research: A Review. JAMA Oncol 2017;3:1722-8.
108 [21] American College of Surgeons. Registry Manuals. https://www.facs.org/ quality-programs/cancer/ncdb/registrymanuals/cocmanuals; 2016, Accessed date: 10 July 2017. [22] American College of Surgeons. National Cancer Database. https://www.facs.org/quality%20programs/cancer/ncdb; 2017, Accessed date: 10 July 2017. [23] Percy C, Fritz A, Ries L. Conversion of neoplasms by topography and morphology from the International Classification of Diseases for Oncology, Second Edition (ICD-O-2) to the International Classification of Diseases for Oncology, Third Edition (ICD-O-3). https://seer.cancer.gov/tools/ conversion/ICDO2-3manual.pdf; 2001, Accessed date: 15 July 2017. [24] American College of Surgeons. Cancer Program Categories. https:// www.facs.org/quality-programs/cancer/coc/apply/categories, Accessed date: 15 March 2016. [25] National Cancer Institute. NCI-Designated Cancer Centers. http://grants.nih.gov/grants/guide/pa-files/PAR-13-386.html, Accessed date: 15 July 2017. [26] Shuch B, Hofmann JN, Merino MJ, et al. Pathologic validation of renal cell carcinoma histology in the Surveillance, Epidemiology, and End Results program. Urol Oncol 2014;32[23 e29-13]. [27] Lipworth L, Morgans AK, Edwards TL, et al. Renal cell cancer histological subtype distribution differs by race and sex. BJU Int 2016;117:260-5. [28] Olshan AF, Kuo TM, Meyer AM, Nielsen ME, Purdue MP, Rathmell WK. Racial difference in histologic subtype of renal cell carcinoma. Cancer Med 2013;2:744-9.
T. Gansler et al. [29] Purdue MP, Moore LE, Merino MJ, et al. An investigation of risk factors for renal cell carcinoma by histologic subtype in two case-control studies. Int J Cancer 2013;132:2640-7. [30] Griswold ME, Swihart BJ, Caffo BS, Zeger SL. Practical marginalized multilevel models. Stat 2013;2:129-42. [31] Amin MB, Epstein JI, Ulbright TM, et al. Best practices recommendations in the application of immunohistochemistry in urologic pathology: report from the International Society of Urological Pathology consensus conference. Am J Surg Pathol 2014;38:1017-22. [32] Reuter VE, Argani P, Zhou M, Delahunt B, Members of the ISUP Immunohistochemistry in Diagnostic Urologic Pathology Group. Best practices recommendations in the application of immunohistochemistry in the kidney tumors: report from the International Society of Urologic Pathology consensus conference. Am J Surg Pathol 2014;38:e35-9. [33] Cheng L, Amin MB, Beltran-Lopez A, Montironi R. Unclassified renal cell carcinoma. In: Moch H, Humphrey PA, Ulbright TM, Reuter VE, editors. World Health Organization Classification of Tumours. Pathology and Genetics of Tumours of the Urinary System and Male Genital Organs. Lyon, France: IARC Press; 2016. p. 41. [34] Berwick DM. Disseminating innovations in health care. JAMA 2003; 289:1969-75. [35] Gansler T, Fedewa SA, Lin CC, Amin MB, Jemal A, Ward EM. Trends in diagnosis of gleason score 2 through 4 prostate cancer in the National Cancer Database, 1990-2013. Arch Pathol Lab Med 2017; 141:1686-96.
www.elsevier.com/locate/humpath
Original contribution
Trends in reporting histological subtyping of renal cell carcinoma: association with cancer center type☆,☆☆ Ted Gansler MD,MBA,MPH a,⁎, Stacey Fedewa PhD, MPH a , Mahul B. Amin MD b , Chun Chieh Lin MBA, PhD a , Ahmedin Jemal DVM, PhD a a
Intramural Research, American Cancer Society,Atlanta, GA 30303 Department of Pathology and Laboratory Medicine, The University of Tennessee Health Science Center, Memphis, TN 38163
b
Received 20 October 2017; revised 3 January 2018; accepted 5 January 2018
Keywords: Renal cell carcinoma; Pathology; Classification; Registries; Healthcare quality
Summary Histological classification of renal cell carcinoma (RCC) has become increasingly important for clinical management. We identified 295 483 RCC diagnosed from 1998-2014 in the National Cancer Database (NCDB) to examine temporal trends in proportions of RCC with unspecified histology and several specific histologies from the 1998 and 2004 World Health Organization classifications of RCC. Further, multivariable log binomial analysis of 101 062 RCC diagnosed from 2010 to 2014 was used to determine whether the association of diagnosing/treating facility type and the proportion of unspecified RCC is independent of patient demographic and clinical factors. Between 1998 and 2014, the proportion of histologically unspecified RCC decreased substantially in all facility types, with the decrease smallest in community programs (from 86.0% to 28.1%) and largest in National Cancer Institute–designated centers (from 85.1% to 9.8%). These decreases were offset by increases in percentages of papillary, clear cell, and chromophobe RCC cases. During 2010 to 2014, relative to community programs, RCCs were 21% less likely to be reported as unspecified histology (adjusted prevalence ratio [aPR] = 0.79; 95% CI, 0.68-0.92) in comprehensive community programs, 32% less likely in integrated network programs (aPR = 0.68; 95% CI, 0.57-0.92) and academic programs (aPR = 0.68; 95% CI, 0.54-0.87), and 63% less likely (aPR = 0.37; 95% CI, 0.26-0.52) in National Cancer Institute –designated programs. These results have implications for the optimal selection of targeted systemic therapies for patients with advanced disease, and for the potential value of cancer registry data in pathology quality improvement programs to promote more rapid and consistent adoption of new classifications of RCC and other neoplasms. © 2018 Elsevier Inc. All rights reserved.
☆
Financial disclosures: None. This study used the National Cancer Data Base (NCDB), which is supported by the American College of Surgeons and the American Cancer Society. The authors acknowledge the efforts of the American College of Surgeons and the Commission on Cancer in the creation of the National Cancer Data Base. The data used in the study are derived from the limited data set of the NCDB. The American College of Surgeons and the Commission on Cancer have not verified and are not responsible for the analytic or statistical methodology employed, or the conclusions drawn from these data by the authors. ⁎ Corresponding author. E-mail addresses: [email protected] (T. Gansler), [email protected] (S. Fedewa), [email protected] (M. B. Amin), [email protected] (C. C. Lin), [email protected] (A. Jemal). ☆☆
https://doi.org/10.1016/j.humpath.2018.01.010 0046-8177/© 2018 Elsevier Inc. All rights reserved.
100
1. Introduction Classification of renal cell carcinoma (RCC) has become substantially more complex during the past two decades, with recognition of several new types of RCC based on progress in understanding the relationships among molecular pathology, morphology, and natural history. For more than a decade, National Comprehensive Cancer Network (NCCN) treatment guidelines have recommended that RCC histology be considered in selecting systemic molecularly targeted therapies [1], and the increasing number of drugs targeting specific molecular alterations present in subsets of RCC has further increased the importance of precise morphologic and molecular classification [2,3]. Because several RCC types are associated with inherited syndromes, precise histological characterization can also help identify families most likely to benefit from genetic counseling and testing [4-8]. In addition, some retrospective studies have noted prognostic differences among certain RCC subtypes, although these differences may not persist after consideration of other variables such as stage and grade [9-14]. Relatively little is known regarding the promptness with which the 1997 Heidelberg classification [15], the 1998 World Health Organization (WHO) classification [16], and the 2004 WHO classification [17] were adopted, and even less is known about the consistency of adopting this new terminology in various practice settings. This information could help in facilitating consistent use of new diagnostic categories introduced by the 2016 WHO classification of RCC and the 2013 International Society of Urological Pathology Vancouver classification [5-8] and, more generally, in quality measurement and quality improvement for other pathology-based disease classification systems. In a previous study using the National Cancer Database (NCDB), we observed that relative to National Cancer Institute (NCI)–designated comprehensive programs, community programs were significantly more likely to record broader diagnoses (rather than narrower diagnoses) for cancers of several organs, and this association was particularly prominent for malignant neoplasms of the kidneys [18]. In the current study, we use the NCDB records to examine trends from 1998 through 2014 in the adoption by various cancer center categories of several of the most common RCC types in the 1997 Heidelberg, 1998 WHO, and 2004 WHO classifications. We also examined whether cancer center categories are independently associated with the prevalence of reporting histologically unspecified RCC during a period (2010-2014) when consideration of histological type is recommended in the selection of targeted therapies for patients with metastatic RCC. Unlike other databases such as the NCI’s Surveillance, Epidemiology, and End Results Program, the NCDB provides information regarding the treatment center category and whether the patient was diagnosed at the treating facility [19-22].
T. Gansler et al.
2. Materials and methods 2.1. Study participants We examined records in the NCDB of patients diagnosed with RCC from 1998 through 2014. The NCDB, a collaboration of the American College of Surgeons and the American Cancer Society, collects data from more than 1 450 Commission on Cancer (CoC)–approved hospitals and approximately 70% of all cancers of the kidneys and renal pelvis diagnosed in the United States [19]. The NCDB contains data on patient demographics, clinical characteristics, and facility type that are abstracted from medical records and recorded according to a standardized data dictionary [20-22]. The Morehouse University Institutional Review Board in Atlanta, Georgia determined that this study is exempt from review. The International Classification of Diseases for Oncology version 3 (ICD-O-3) was first used by the NCDB in 2001 and although the International Classification of Diseases for Oncology version 2 (ICDO-2) included codes for papillary RCC and clear cell RCC, there was no code in the ICDO-2 for chromophobe RCC, cyst-associated RCC, sarcomatoid RCC, and renal collecting duct carcinoma [23]. By starting analyses in 1998, we were therefore able to include baseline data before adoption of ICDO-3. We limited some analyses to cases diagnosed from 2010 through 2014 to reflect patterns of pathology interpretation during the current era in which RCC histologic types are recognized by oncology guidelines as a factor in choosing systemic therapy for patients with metastatic RCC [1]. We initially selected patients diagnosed and reported by the same CoC-accredited facility, with a malignant neoplasm of the kidney confirmed by histology, and with ICD-O-3 histology codes 8260 (papillary adenocarcinoma, NOS), 8310 (clear cell adenocarcinoma, NOS), 8312 (renal cell carcinoma), 8316 (cyst-associated renal cell carcinoma), 8317 (renal cell carcinoma, chromophobe type), 8318 (renal cell carcinoma, sarcomatoid), 8319 (collecting duct carcinoma), or 8320 (granular cell carcinoma). Although there are several other types of RCC in the 2004 WHO classification, such as renal carcinomas associated with Xp11.2 translocation/TFE3 gene fusions, RCC associated with neuroblastoma, and mucinous tubular and spindle cell carcinoma, there are no specific ICDO-3 codes to represent these histologies; therefore, these cases could not be identified in NCDB records. Thus, the ICD-O-3 histology code, 8312, includes (1) RCC cases diagnosed only as “RCC” by the pathologist but which actually had specific histologies with corresponding ICDO-3 codes (such as papillary, clear cell, and chromophobe), (2) a much smaller number of cases with rare RCC histologies that may have been diagnosed by the pathologist but which were assigned a more general code by the registry because there are no specific ICDO-3 codes for these entities, and (3) a small number of cancers meeting criteria for unclassified RCC, a diagnosis for which there is no corresponding ICDO-3 code.
RCC histology trends Table 1
101
Exclusion criteria for renal cell carcinoma cases, National Cancer Database 1998-2014 and 2010-2014
Included and excluded cases
No. 1998-2014
No. 2010-2014
Initial cases of biopsy-confirmed renal cell carcinoma, diagnosed and reported at same CoC-accredited facility Exclude missing sex Exclude missing race Exclude age b20 and age ≥90 Exclude missing AJCC stage group stage or stage 0 Exclude missing summary surgery of primary site or no nephrectomy Exclude “other” or pediatric facility type Final
385542
130093
95 25788 442 + 1132 26107 + 3 134 + 45731 972 295483 (76.6%)
22 4894 146 + 409 7687 + 1 39 + 18473 200 101062 (77.7%)
NOTE. Some cases had more than one exclusion criterion, so subtracting individual exclusions from the initial cases does not exactly equal the final sample. Abbreviations: AJCC, American Joint Committee on Cancer; CoC, Commission on Cancer.
2.2. Outcome and independent variables The primary outcome was histologic type of RCC. Because most histologic types of RCC are too rare for sufficiently precise analyses even in this large data set, we focused on histologically unspecified RCC (coded as RCC without any specific type), and the 3 most common specific types—papillary, clear cell, and chromophobe RCC. Although sarcomatoid RCCs were included among the histologies for including RCC cases, the Table 2
diagnosis of sarcomatoid RCC was not an outcome in our analyses because the Heidelberg classification and the 2004 WHO classification do not consider it as a specific type of RCC, but instead as a secondary change that can occur in RCC with various histologic types (for example, clear cell RCC with focal sarcomatoid change). Similarly, granular RCC was included in the denominator for RCC cases, but not individually reported as an outcome because it is considered an obsolete diagnosis and is not included in the Heidelberg classification and the
Patient characteristics by facility type among renal cell carcinoma cases, National Cancer Database 1998-2014 CP, n (%)
Year group (P b .001) 1998-2000 3653 (14.8) 2001-2005 7686 (31.2) 2006-2010 7616 (30.9) 2011-2014 5674 (23.0) Age group (P b .001) 20-39 years 1005 (4.1) 40-49 years 2985 (12.1) 50-59 years 5873 (23.8) 60-69 years 7120 (28.9) 70-79 years 5715 (23.2) 80-89 years 1932 (7.8) Sex (P b .001) Male 14965 (60.8) Female 9665 (39.2) Race/ethnicity (P b .001) Non-Hispanic white 20337 (82.6) Hispanic 1369 (5.6) Black 2249 (9.1) Asian and PI 406 (1.6) Other 269 (1.1) AJCC stage group (P b .001) I 15690 (63.7) II 3359 (13.6) III 3862 (15.7) IV 1719 (7.0) Total 24630
CCP, n (%)
INP, n (%)
ACADP, n (%)
NCIP, n (%)
All Facilities, n (%)
15400 (12.4) 35319 (28.4) 40538 (32.6) 33030 (26.6)
4418 (12.4) 10420 (29.3) 11671 (32.9) 9016 (25.4)
8292 (11.1) 19492 (26.1) 23936 (32.1) 22955 (30.7)
3648 (10.0) 9703 (26.7) 11504 (31.6) 11511 (31.7)
35411 (12.0) 82620 (28.0) 96266 (32.2) 82186 (27.8)
5296 (4.3) 15209 (12.2) 30465 (24.5) 36167 (29.1) 28473 (22.9) 8677 (7.0)
1694 (4.8) 4811 (13.5) 9131 (25.7) 10295 (29.0) 7525 (21.2) 2069 (5.8)
4322 (5.8) 11026 (14.8) 20072 (26.9) 21157 (28.3) 14323 (19.2) 3775 (5.1)
2189 (6.0) 5354 (14.7) 9670 (26.6) 10924 (30.0) 6756 (18.6) 1473 (4.1)
14506 (4.9) 39385 (13.3) 75211 (25.5) 85663 (29.0) 62792 (21.3) 17926 (6.1)
76475 (61.5) 47812 (38.5)
21981 (61.9) 13544 (38.1)
46247 (61.9) 28428 (38.1)
23638 (65.0) 12728 (35.0)
183306 (62.0) 112177 (38.0)
103170 (83.0) 6676 (5.4) 11589 (9.3) 1617 (1.3) 1235 (1.0)
26929 (75.8) 2542 (7.2) 5252 (14.8) 428 (1.2) 374 (1.1)
54424 (72.9) 5362 (7.2) 12127 (16.2) 1621 (2.2) 1141 (1.5)
28876 (79.4) 1630 (4.5) 4591 (12.6) 790 (2.2) 479 (1.3)
233736 (79.1) 17579 (5.9) 35808 (12.1) 4862 (1.6) 3498 (1.2)
82188 (66.1) 16199 (13.0) 18486 (14.9) 7414 (6.0) 124287
23730 (66.8) 4348 (12.1) 5409 (15.2) 2038 (5.7) 35525
50533 (67.7) 8371 (11.2) 11248 (15.5) 4523 (6.1) 74675
25079 (69.0) 3367 (9.3) 5653 (15.5) 2267 (6.2) 36366
197220 (66.7) 35644 (12.1) 44658 (15.1) 17961 (6.1) 295483
NOTE. The χ2 tests and the corresponding two-tailed P-values represent differences in facility type by patient characteristics. Abbreviations: CP, Community Cancer Program; CCP, Comprehensive Community Cancer Program; INP, Integrated Network Cancer Program; ACADP, Academic Cancer Program; NCIP, National Cancer Institute-Designated Cancer Program; PI, Pacific Islander.
102
T. Gansler et al.
2004 WHO classification [15-17]. Collecting duct carcinoma diagnoses were not studied because there were too few cases for precise analysis. For some analyses, we dichotomized the histology variable to unspecified RCC versus any specific RCC histology (excluding, as noted above, sarcomatoid and granular RCC). Our primary independent variable of interest was facility type. Five facility types were studied—community programs, comprehensive community programs, integrated network programs, academic programs, and NCI-designated programs. Criteria for cancer center classification are described in detail elsewhere [24,25]. Patients reported by Pediatric programs and “other” program types were excluded because of small numbers. Briefly, the spectrum from community programs to comprehensive community programs and integrated network programs, to academic programs, and to NCI-designated programs generally reflects greater comprehensiveness of diagnostic and treatment services, numbers of patients treated, participation in clinical research, and involvement in postgraduate training of physicians. Our interest in this variable is based largely on our prior study of cases diagnosed prior to widespread use of targeted therapies for metastatic RCC, showing a higher percentage of unspecified RCC cases reported by community programs and comprehensive community programs Table 3
relative to academic programs and NCI-designated programs [18]. This interest is supported by prior reports of differences in RCC histology interpretation between records from cancer registries and review of the same specimens by subspecialists in renal or urological pathology [26]. Also, the percentage of cases recorded only as “RCC” or as “RCC not otherwise specified” is much higher in reports based on cancer registries (which include cases from all facility types) than in series of cases from individual academic cancer centers or groups of several academic centers [10,14,27-29]. We also examined variables that might potentially be associated with histology and facility type, such as year of diagnosis, patient age and sex, patient race/ethnicity, and American Joint Committee on Cancer (AJCC) stage group; cases missing values for these variables were therefore excluded as shown in Table 1. The stage group recorded in the NCDB for each patient was based on the AJCC version that was in effect at the time of diagnosis; the 7th edition was used 20102014, the 6th edition staging was used in 2003-2009, and 5th edition staging was used in 1998-2002. Patients younger than 20 years and those older than 89 years were excluded because of small numbers in these groups. Finally, patients who did not undergo nephrectomy (partial, total, radical, or unspecified)
Patient characteristics by facility type among renal cell carcinoma cases, National Cancer Database 2010 to 2014 CP, n (%)
Year (P b .001) 2010 1446 (20.3) 2011 1390 (19.5) 2012 1449 (20.4) 2013 1367 (19.2) 2014 1468 (20.6) Age group (P b .001) 20-39 years 315 (4.4) 40-49 years 863 (12.1) 50-59 years 1768 (24.8) 60-69 years 2179 (30.6) 70-79 years 1530 (21.5) 80-89 years 465 (6.5) Sex (P b .001 Male 4428 (62.2) Female 2692 (37.8) Race/ethnicity (P b .001) Non-Hispanic white 5631 (79.1) Hispanic 509 (7.1) Black 747 (10.5) Asian & PI 138 (1.9) Other 95 (1.3) AJCC stage group (P b .001) I 4775 (67.1) II 900 (12.6) III 1041 (14.6) IV 404 (5.7) Total 7120
CCP, n (%)
INP, n (%)
ACADP, n (%)
NCIP, n (%)
All Facilities, n (%)
7820 (19.1) 7952 (19.5) 8195 (20.1) 8477 (20.8) 8406 (20.6)
2167 (19.4) 2313 (20.7) 2279 (20.4) 2188 (19.6) 2236 (20.0)
4975 (17.8) 5549 (19.9) 5742 (20.6) 5822 (20.8) 5842 (20.9)
2468 (17.7) 2601 (18.6) 2987 (21.4) 2931 (21.0) 2992 (21.4)
18876 (18.7) 19805 (19.6) 20652 (20.4) 20785 (20.6) 20944 (20.7)
1897 (4.6) 4834 (11.8) 10298 (25.2) 12794 (31.3) 8537 (20.9) 2490 (6.1)
567 (5.1) 1482 (13.3) 2909 (26.0) 3459 (30.9) 2238 (20.0) 528 (4.7)
1686 (6.0) 4079 (14.6) 7626 (27.3) 8442 (30.2) 4896 (17.5) 1201 (4.3)
875 (6.3) 1965 (14.1) 3831 (27.4) 4434 (31.7) 2394 (17.1) 480 (3.4)
5340 (5.3) 13223 (13.1) 26432 (26.2) 31308 (31.0) 19595 (19.4) 5164 (5.1)
25345 (62.0) 15505 (38.0)
7054 (63.1) 4129 (36.9)
17448 (62.5) 10482 (37.5)
9130 (65.3) 4849 (34.7)
63405 (62.7) 37657 (37.3)
33052 (80.9) 2636 (6.5) 4044 (9.9) 610 (1.5) 508 (1.2)
8354 (74.7) 803 (7.2) 1744 (15.6) 152 (1.4) 130 (1.2)
19364 (69.3) 2353 (8.4) 4888 (17.5) 765 (2.7) 560 (2.0)
10864 (77.7) 734 (5.3) 1856 (13.3) 311 (2.2) 214 (1.5)
77265 (76.5) 7035 (7.0) 13279 (13.1) 1976 (2.0) 1507 (1.5)
28595 (70.0) 4520 (11.1) 5779 (14.1) 1956 (4.8) 40850
7834 (70.1) 1189 (10.6) 1670 (14.9) 490 (4.4) 11183
19749 (70.7) 2575 (9.2) 4280 (15.3) 1326 (4.7) 27930
9973 (71.3) 1098 (7.9) 2243 (16.0) 665 (4.8) 13979
70926 (70.2) 10282 (10.2) 15013 (14.9) 4841 (4.8) 101062
NOTE. The χ2 tests and the corresponding 2-tailed P-values represent differences in facility type by patient characteristics. Abbreviations: CP, Community Cancer Program; CCP, Comprehensive Community Cancer Program; INP, Integrated Network Cancer Program; ACADP, Academic Cancer Program; NCIP, National Cancer Institute–Designated Cancer Program; PI, Pacific Islander.
RCC histology trends
103
were excluded because diagnosis of these cases as unspecified RCC might have been influenced by the amount of tissue available (especially with core biopsies), and our intent was to examine factors entirely under the pathologist’s control.
2.3. Data analysis We performed bivariate statistical analyses with SPSS Statistics version 24 (Armonk, NY). χ2 Tests and the corresponding two-tailed P-values with a significance level at .05 were used to examine potential associations of RCC histology with the independent variables noted above. Multivariable logbinomial analyses were performed with SAS, version 9.3 (SAS, Cary, NC). Adjusted multivariable log binomial models, which accounted for clustering at the facility level (correlation of characteristics among patients treated at a particular hospital) were used to estimate adjusted prevalence ratios (aPR) and 95% confidence intervals (CI) for unspecified RCC versus all other RCC types combined [30].
3. Results 3.1. Study participants and characteristics, 1998-2014 Of the initial 385,542 cases of RCC recorded in the NCDB from 1998 through 2014, 295,483 (76.6%) remained in the
analytic sample after exclusions; whereas 101 062 (77.7%) of the 130 093 cases diagnosed from 2010 through 2014 remained after exclusions (Table 1). Characteristics of the analytic sample for RCC cases diagnosed from 1998 through 2014 and for the subset diagnosed from 2010 through 2014 are summarized in Tables 2 and 3, respectively. During both time periods, the modal age category was 60 to 69 years, a majority of patients were male, and more than three fourths were non-Hispanic whites. Approximately two-thirds of patients had AJCC stage I disease. There were statistically significant associations between facility type and all-patient sociodemographic and clinical characteristics, but the magnitudes of these differences generally were small except for patient age.
3.2. Trends in percentages of RCC with specific histologies and unspecified histology, 1998-2014 The percentage of histologically unspecified RCC steadily decreased between 1998 and 2014, with corresponding increases in percentages of papillary, clear cell, and chromophobe RCC cases (Figure). Unspecified RCC comprised 86.0% of all RCC diagnosed by community programs in 1998, but only 28.1% in 2014. This decline was even more prominent in NCI-designated programs (85.1% in 1998 versus 9.8% in 2014), and was intermediate for other facility categories (for example, 86.8% in 1998 versus 15.4 in 2014 in integrated network programs).
25
70 60
% Clear Cell RCC
% Papillary RCC
20
15
10
50
40 30 20
5 10 0 1998
2000 CP
2002
2004
CCP
2006 INP
2008
2010
ACADP
2012
0 1998
2014
% RCC with Unspecified Histology
9
% Chromophobe RCC
2002
2004
CCP
2006 INP
2008
2010
ACADP
2012
2014
NCIP
100
10
8 7 6 5 4 3 2 1 0 1998
2000 CP
NCIP
2000 CP
2002 CCP
2004
2006 INP
2008
2010
ACADP
2012 NCIP
2014
90 80 70 60 50 40 30 20 10 0 1998
2000 CP
2002 CCP
2004
2006 INP
2008
2010
ACADP
2012
2014
NCIP
Figure Time trends in percentage of renal cell carcinoma cases by histologic type. Abbreviations: ACADP, Academic (teaching/research) Cancer Programs; CCP, Comprehensive Community Cancer Programs; CP, Community Cancer Programs; INP, Integrated Network Cancer Programs; NCIP, National Cancer Institute-designated Cancer Programs, RCC, renal cell carcinoma.
104
T. Gansler et al.
For all years from 2002 through 2014, NCI-designated programs diagnosed the smallest percentage of RCC cases with unspecified histology, and the largest percentages of papillary, clear cell, and chromophobe RCC cases. Conversely, community programs diagnosed the largest percentage of RCC cases with unspecified histology and the smallest percentage of chromophobe RCC and papillary RCC for all years during this interval (albeit, tied with comprehensive community programs during 2012 for the lowest percentage of papillary RCC), as well as the smallest percentage of clear cell RCC during 10 of these 13 years. During all years from 2002 through 2014, academic programs had the second or third lowest percentages of RCC
Table 4
with unspecified histology. Academic centers had the second or third highest percentages during each of these years for papillary and chromophobe RCC, and during 10 of these 13 years for clear cell RCC. A graph for granular RCC is not shown in the Figure because this term is not used in the 2004 WHO classification. As expected, however, we observed a steady decline in use of this term, from 1.8% of all RCC diagnoses during 1998 to 2000, to 0.1% during 2011 through 2014. Overall, this diagnosis was used least often in NCI-designated programs during this period (0.4%) and most often (0.8%) by community programs, comprehensive community programs, and integrated network programs (not shown).
Patient, tumor, and facility characteristics by renal cell carcinoma histology, National Cancer Database 2010 to 2014 Papillary N (%)
Year (P b.001) 2010 2684 (14.2) 2001 2921 (14.7) 2012 3183 (15.4) 2013 3183 (15.3) 2014 3170 (15.1) Age group (P b.001) 20-39 years 519 (9.7) 40-49 years 1609 (12.2) 50-59 years 3898 (14.7) 60-69 years 5131 (16.4) 70-79 years 3208 (16.4) 80-89 years 776 (15.0) Sex (P b.001) Male 11362 (17.9) Female 3779 (10.0) Race/ethnicity (P b.001) Non-Hispanic white 9880 (12.8) Hispanic 451 (6.4) Black 4500 (33.9) Asian & PI 182 (9.2) Other 128 (8.5) AJCC stage group (P b.001) I 11902 (16.8) II 1596 (15.5) III 1341 (8.9) IV 302 (6.2) Facility type (P b.001) CP 841 (11.8) CCP 5216 (12.8) INP 1805 (16.1) ACADP 4658 (16.7) NCIP 2621 (18.7) Total 15141 (15.0)
Clear cell (%)
Chromophobe N (%)
Any specific type N (%)
Unspecified N (%)
10331 (54.7) 11128 (56.2) 11679 (56.6) 12099 (58.2) 12399 (59.2)
1079 (5.7) 1177 (5.9) 1274 (6.2) 1310 (6.3) 1309 (6.3)
14260 (75.6) 15409 (77.8) 16288 (78.9) 16753 (80.6) 17022 (81.3)
4348 (23.0) 4167 (21.0) 4143 (20.1) 3799 (18.3) 3711 (17.7)
3048 (57.1) 7729 (58.5) 15250 (57.7) 17731 (56.6) 10984 (56.1) 2894 (56.0)
537 (10.1) 946 (7.2) 1521 (5.8) 1639 (5.2) 1179 (6.0) 327 (6.3)
4167 (78.0) 10429 (78.9) 20876 (79.0) 24732 (79.0) 15505 (79.1) 4023 (77.9)
1138 (21.3) 2664 (20.1) 5239 (19.8) 6213 (19.8) 3848 (19.6) 1066 (20.6)
35045 (55.3) 22591 (60.0)
3452 (5.4) 2697 (7.2)
50340 (79.4) 29392 (78.1)
12285 (19.4) 7883 (20.9)
46093 (59.7) 4702 (66.8) 4688 (35.3) 1270 (64.3) 883 (58.6)
4627 (6.0) 418 (5.9) 862 (6.5) 137 (6.9) 105 (7.0)
61138 (79.12) 5614 (79.8) 10258 (77.3) 1599 (80.9) 1123 (74.5)
15248 (19.7) 1344 (19.1) 2863 (21.6) 353 (17.9) 360 (23.9)
40184 (56.7) 5351 (52.0) 9256 (61.7) 2845 (58.5)
4256 (6.0) 1055 (10.3) 770 (5.1) 68 (1.4)
56948 (80.3) 8062 (78.4) 11450 (76.3) 3273 (67.6)
13715 (19.3) 2106 (20.5) 3212 (21.4) 1135 (23.4)
3782 (53.1) 23585 (57.7) 6243 (55.8) 15507 (55.5) 8519 (60.9) 57636 (57.0)
321 (4.5) 2218 (5.4) 725 (6.5) 1784 (6.4) 1101 (7.9) 6149 (6.1)
4982 (70.0) 31300 (76.6) 8875 (79.4) 22174 (79.4) 12401 (88.7) 79732 (78.9)
2024 (28.4) 9087 (22.2) 2168 (19.4) 5439 (19.5) 1450 (10.4) 20168 (20.0)
NOTE. The χ2 tests and the corresponding 2-tailed P values represent differences between any specific histological type and unspecified RCC by patient characteristics. The “any specific type” category includes papillary, clear cell, and chromophobe RCC as well as collecting duct carcinoma, which is not shown separately because of too few cases. The total of “any specific type” and “unspecified” is slightly less than 100% because neither group includes granular RCC or sarcomatoid RCC. Please refer to the Methods section for a complete explanation. Abbreviations: AJCC, American Joint Committee on Cancer; CP, Community Cancer Program; CCP, Comprehensive Community Cancer Program; INP, Integrated Network Cancer Program; ACADP, Academic Cancer Program; NCIP, National Cancer Institute–Designated Cancer Program; PI, Pacific Islander.
RCC histology trends
3.3. Bivariate associations with RCC histology, 2010-2014 From 2010 through 2014, 79 732 cases (78.9%) were reported as a specific RCC type (excluding, as noted previously, 1162 cases with granular or sarcomatoid histologies), whereas 20168 (20.0%) were reported as histologically unspecified RCC (Table 4). Aided by this very large sample size, all independent variables were very significantly (P b .001) associated with differences in proportions of RCC with any specific histology versus unspecified RCC. However, the magnitudes of these associations seemed clinically significant for only a few of these variables. The percentages of unspecified RCC cases decreased from 23.0% in 2010 to 17.7% in 2014, with corresponding increases in the percentages of papillary, clear cell, and chromophobe RCC cases. Relative to males, females were much less often diagnosed with papillary RCC and more likely to have clear cell or chromophobe RCC. Blacks were by far the most likely (33.3%) to be diagnosed with papillary RCC, whereas this histology was least common among Hispanics (6.4%). Conversely, clear cell RCC was least common among Blacks (35.3%) and most common among Hispanics (66.8%). Stage IV RCC cases were least likely (67.6%) to be assigned a specific histology, and stage I RCCs were the most likely (80.3%), primarily due to different percentages of papillary RCC. RCC cases diagnosed in NCI-designated programs were the least likely (10.4%) and those from community programs the most likely (28.4%) to have unspecified histology. In a supplemental analysis (full results not shown) of RCC cases that were diagnosed and classified based on a biopsy specimen and which had not undergone any surgical treatment (but which otherwise met the same inclusion criteria of our main analysis), the percentages of histologically unspecified RCC cases from various facility types showed a similar pattern, with the main difference being consistently larger proportion of unspecified RCC cases. Biopsy-only cases from community programs were the most likely (57.0% of 1042) to have unspecified histology; those from NCIdesignated programs were the least likely (32.0% of 993); other facility categories had intermediate values (comprehensive community programs, 51.7% of 4079; integrated network programs, 49.0% of 940; and academic programs, 50.1%of 2209).
3.4. Independent associations with percentages of unspecified RCC, 2010-2014 A log-binomial model, accounting for clustering at the hospital level and including potential confounding covariates such as year of diagnosis, age, sex, race/ethnicity, and AJCC stage group demonstrated an independent, statistically significant, and clinically significant association between facility type and the proportion of RCC cases with unspecified histology (Table 5). Relative to community programs, the prevalence of a diagnosis of unspecified RCC in other facility types was significantly lower, with aPRs of 0.79 (95% CI, 0.68-0.92)
105 Table 5 Adjusted prevalence ratio of histologically unspecified renal cell carcinoma, National Cancer Database 2010-2014
Year 2010 2011 2012 2013 2014 Age group 20-39 years 40-49 years 50-59 years 60-69 years 70-79 years 80-89 years Sex Male Female Race/Ethnicity Non-Hispanic white Hispanic Black Asian & PI Other AJCC stage group I II III IV Facility type CP CCP INP ACADP NCIP
Prevalence ratio
95% Confidence interval
P
1.28 1.18 1.14 1.03 referent
1.17-1.41 1.09-1.28 1.05-1.23 0.97-1.09
b.001 b.001 b.001 .362
1.12 1.05 1.02 1.01 0.99 referent
1.02-1.23 0.97-1.13 0.95-1.09 0.95-1.08 0.93-1.06
.014 .255 .610 .663 .760
Referent 1.07
1.04-1.10
b.001
referent 0.94 1.12 0.93 1.23
0.82-1.08 1.04-1.21 0.74-1.18 1.07-1.40
.40 .002 .57 .003
0.77-0.88 0.79-0.91 0.87-1.00
b.001 b.001 .05
0.68-0.92 0.57-0.92 0.54-0.87 0.26-0.52
.003 .002 b.001 b.001
0.82 0.85 0.93 referent referent 0.79 0.68 0.68 0.37
NOTE. Results are based on a multivariable log-binomial model. Abbreviations: AJCC, American Joint Committee on Cancer; CP, Community Cancer Program; CCP, Comprehensive Community Cancer Program; INP, Integrated Network Program; ACADP, Academic Cancer Program; NCIP, National Cancer Institute–Designated Cancer Program; PI, Pacific Islander.
for comprehensive community programs, 0.68 for integrated network programs (95% CI, 0.57-0.92), 0.68 for academic programs (95% CI, 0.54-0.87), and most notably, 0.37 (95% CI, 0.26-0.52) for NCI-designated programs. Although most values for all covariates were significantly different from the referent categories, their effect sizes were much smaller than those for facility type.
4. Discussion This study of NCDB records demonstrated a dramatic decline between 1998 and 2014 in the percentage of RCC cases
106 recorded with unspecified histology, accompanied by inverse trends in the percentages of papillary, clear cell, and chromophobe RCC cases. These trends were most prominent and occurred earliest at NCI-designated cancer programs and latest at community programs. An adjusted analysis of RCC cases diagnosed from 2010 through 2014 demonstrated that facility type was strongly associated with the likelihood of RCC being diagnosed as having unspecified histology, independent of patient demographic and clinical factors. Despite recommendations from the 1997 Heidelberg [15], the 1998 WHO [16] and the 2004 WHO classifications [17] that specific histological types including clear cell, papillary, and chromophobe RCC be reported as such rather than as unspecified RCC, adoption occurred slowly and inconsistently. One likely reason for the persistently high percentage of histologically unspecified cases during this period is that until recently, although specific histological diagnoses had implications for prognosis and for genetic counseling of some patients [4-14], they did not influence treatment decisions. However, beginning approximately in 2007 [1], recommendations for molecularly targeted therapies for metastatic RCC varied among some specific RCC types, so it is somewhat surprising that 20% of RCC cases were unspecified during the period from 2010 through 2014. This percentage seems even more surprising given that most clear cell, papillary, and chromophobe RCCs can be recognized on routine hematoxylin and eosin–stained sections, and that in cases with equivocal histology, subtyping can usually be accomplished by the use of readily available techniques such as immunohistochemistry and fluorescence in situ hybridization [8,31,32]. Although the NCDB does not contain information on each center’s access to more sophisticated molecular testing or use of such testing for individual cases, our interpretation of these results is that the differences noted above primarily reflect variations in pathologists’ interpretation and classification of specimens processed using widely available technology. The 2016 WHO classification notes that unclassified RCC is a cancer with morphological, immunohistochemical, and chromosomal/molecular characteristics that do not correspond to any of the known common, rare, and recently described types of RCC, or to other cancers in the kidneys, such as urothelial carcinoma or metastatic cancer. When this definition is applied, only 5% to 7% of RCC should remain unclassified, a proportion much smaller than the 20% of histologically unspecified RCC in this study from 2010 through 2014 [33]. Based on the substantial difference between these two percentages, we conclude that the majority of histologically unspecified RCC do not meet WHO criteria for diagnosis as unclassified RCC. The gradual and inconsistent adoption of the 1997 Heidelberg classification and the 1998 and 2004 WHO classifications seems consistent with the literature on the “diffusion of innovation” in healthcare [34]. The consensus of published evidence, albeit largely limited to adoption of therapeutic interventions rather than diagnostic classifications, suggests that earlier adoption of medical innovations occurs in academic/research/referral institutions in contrast to community
T. Gansler et al. settings, and that complete and consistent adoption often requires several years. This pattern is consistent with our observation that the transition from “lumping” to “splitting” in RCC diagnosis occurred several years earlier and occurred most completely in NCI-designated programs relative to community programs, with an intermediate pattern among other facility types. The observational design of this study and its use of secondary data from the NCDB raise some concern regarding confounding by differences in patient demographic and clinical characteristics that might be associated with histology. These differences might arise because of differences in populations served or from selective referral of certain histological types. However, we addressed concerns regarding referral bias by (1) restricting all analyses to cases that were diagnosed and received their initial course of treatment at the same center, thereby excluding patients who were referred based on initial RCC histology and had histology re-classified by the second facility and (2) demonstrating that differences in percentages of histologically unspecified RCC among various facility types are independent of year of diagnosis; patient age, sex, and race; and AJCC stage group. Furthermore, one might expect advanced, aggressive, poorly differentiated RCC to be referred to larger and more specialized centers, but if this referral pattern occurred, the effect would be a larger rather than smaller proportion of RCC cases with unspecified histology. In adjusted models including these covariates, unspecified RCC is diagnosed approximately one third as often in NCIdesignated programs as in community programs (aPR = 0.37). In addition, relative to diagnoses in community programs, the likelihood of an unspecified RCC diagnosis is significantly lower in all other facility types, and approximately one fourth to one-third less common among comprehensive community programs (PR = 0.79), integrated network programs (aPR = 0.68), and academic programs (aPR = 0.68). Furthermore, patients with cancers of some other sites may require treatment that depends on specialized radiotherapy instruments or surgical expertise that is specific to that site and type (and therefore can lead to referral bias that limits analyses of classification according to facility types), but this is not the case with RCC. Estimating the impact, if any, of these diagnostic variations on health outcomes such as survival is beyond the scope of this investigation. However, it seems very possible that inappropriate “lumping” of some cases as unspecified RCC could have presented challenges to appropriate use of molecularly targeted systemic therapies and in identifying families that might benefit from genetic counseling relevant to hereditary cancer syndromes. It is possible that some of the variation in diagnostic categories among facilities and facility types is due to differences in abstraction of pathology reports by facility cancer registries. However, registry quality control procedures that are mandatory for CoC-approved facilities that submit cases to the NCDB most likely limit this effect, if any, to a small fraction of the observed variation. It is also possible that some communication regarding specific RCC
RCC histology trends histologic types might occur between pathologists and the oncologists who choose targeted therapies outside of formal pathology reports (ie, during multidisciplinary case management conferences, and during conversations in person or by telephone). Despite these limitations, we conclude that the results of these analyses are sufficiently robust to justify consideration of use of similar analyses in quality of cancer care measurement and quality improvement programs. However, an important limitation regarding use of registry data for monitoring the adoption of new disease classification systems is that the coding schema used by registries must include codes specific to the new diagnostic categories. Although the IDCO-3 classification used by registries included most of the histological categories in the 2004 WHO classification, it does not include most of the categories introduced by the WHO in 2016. The practical interpretation of these results is, as we have emphasized in prior studies of other cancers, not to denigrate the value of or challenges faced by pathologists in community practice settings [18,35]. Rather, our goal is to demonstrate the value of cancer registry data in identifying variations in practice patterns that might be efficiently and effectively minimized by a combination of feedback to individual centers comparing their data to benchmark values and availability of relevant continuing education programs. In fact, one of our observations in this study of RCC classification (data not shown) and in our previous studies [18,35] of other aspects of oncological pathology practice is that although there are highly significant differences among facility types, there is also prominent variation within each facility category. Thus, these data are relevant to improving consistency of histologic diagnoses for all facility categories. This strategy for quality measurement applies to a limited scope of practice pattern in surgical pathology and is expected to augment rather than replace ongoing programs of pathology organizations. However, as new technologies enhance transfer of data from medical records and laboratory information systems, this strategy might be extended to additional aspects of anatomical pathology practice.
5. Conclusions This study demonstrates substantial variations among cancer center categories in the promptness and completeness with which they utilized several diagnostic entities introduced by the 1997 Heidelberg and the 1998 WHO classifications of RCC. This diagnostic variation persists in the most recent data available (2014) and may limit application of oncology guidelines that recommend consideration of RCC histological type in selecting molecularly targeted drugs. More broadly, these results demonstrate the potential value of cancer registry data (contingent on registry use of a coding schema that reflects current diagnostic terminology) in some aspects of oncological pathology quality measurement.
107
References [1] Motzer TJ, Bolger GB, Boston B, et al. Kidney Cancer Version 2.2007. NCCN Clinical Practice Guidelines in Oncology. nccn.org; 2007. Accessed 2007. [2] Choueiri TK, Motzer RJ. Systemic therapy for metastatic renal-cell carcinoma. N Engl J Med 2017;376:354-66. [3] Motzer RJ, Jonasch E, Agarwal N, et al. Kidney Cancer Version 2.2017. NCCN Clinical Practice Guidelines in Oncology 2017. nccn.org, Accessed date: 17 April 2017. [4] Leung C, Pan S, Shuch B. Management of renal cell carcinoma in young patients and patients with hereditary syndromes. Curr Opin Urol 2016; 26:396-404. [5] Srigley JR, Delahunt B, Eble JN, et al. The International Society of Urological Pathology (ISUP) Vancouver Classification of Renal Neoplasia. Am J Surg Pathol 2013;37:1469-89. [6] Udager AM, Mehra R. Morphologic, molecular, and taxonomic evolution of renal cell carcinoma: a conceptual perspective with emphasis on updates to the 2016 World Health Organization Classification. Arch Pathol Lab Med 2016;140:1026-37. [7] Moch H, Cubilla AL, Humphrey PA, Reuter VE, Ulbright TM. The 2016 WHO Classification of Tumours of the Urinary System and Male Genital Organs—Part A: Renal, Penile, and Testicular Tumours. Eur Urol 2016; 70:93-105. [8] Moch H, Amin MB, Argani P, et al. Renal cell tumours introduction. In: Moch H, Humphrey PA, Ulbright TM, Reuter VE, editors. World Health Organization Classification of Tumours. Pathology and Genetics of Tumours of the Urinary System and Male Genital Organs. Lyon, France: IARC Press; 2016. p. 14-7. [9] Moch H, Gasser T, Amin MB, Torhorst J, Sauter G, Mihatsch MJ. Prognostic utility of the recently recommended histologic classification and revised TNM staging system of renal cell carcinoma: a Swiss experience with 588 tumors. Cancer 2000;89:604-14. [10] Amin MB, Amin MB, Tamboli P, et al. Prognostic impact of histologic subtyping of adult renal epithelial neoplasms: an experience of 405 cases. Am J Surg Pathol 2002;26:281-91. [11] Nese N, Paner GP, Mallin K, Ritchey J, Stewart A, Amin MB. Renal cell carcinoma: assessment of key pathologic prognostic parameters and patient characteristics in 47,909 cases using the National Cancer Data Base. Ann Diagn Pathol 2009;13:1-8. [12] Keegan KA, Schupp CW, Chamie K, Hellenthal NJ, Evans CP, Koppie TM. Histopathology of surgically treated renal cell carcinoma: survival differences by subtype and stage. J Urol 2012;188:391-7. [13] Tilki D, Nguyen HG, Dall'Era MA, et al. Impact of histologic subtype on cancer-specific survival in patients with renal cell carcinoma and tumor thrombus. Eur Urol 2014;66:577-83. [14] Patard JJ, Leray E, Rioux-Leclercq N, et al. Prognostic value of histologic subtypes in renal cell carcinoma: a multicenter experience. J Clin Oncol 2005;23:2763-71. [15] Kovacs G, Akhtar M, Beckwith BJ, et al. The Heidelberg classification of renal cell tumours. J Pathol 1997;183:131-3. [16] Storkel S, Eble JN, Adlakha K, et al. Classification of renal cell carcinoma: Workgroup No. 1. Union Internationale Contre le Cancer (UICC) and the American Joint Committee on Cancer (AJCC). Cancer 1997;80:987-9. [17] Eble JN, Sauter G, Epstein JI, Sesterhenn IA, editors. Pathology and genetics of tumours of the urinary system and male genital organs. World Health Organization Classification of Tumours. Lyon, France: IARC Press; 2004. p. 10. [18] Gansler T, Fedewa SA, Flanders D, Virgo KS, Ward EM. “Lumping” vs “splitting” in oncologic pathology: association with cancer center type and case volume. J Registry Manag 2012;39:43-52. [19] Lerro CC, Robbins AS, Phillips JL, Stewart AK. Comparison of cases captured in the National Cancer Data Base with those in populationbased central cancer registries. Ann Surg Oncol 2013;20:1759-65. [20] Boffa DJ, Rosen JE, Mallin K, et al. Using the National Cancer Database for Outcomes Research: A Review. JAMA Oncol 2017;3:1722-8.
108 [21] American College of Surgeons. Registry Manuals. https://www.facs.org/ quality-programs/cancer/ncdb/registrymanuals/cocmanuals; 2016, Accessed date: 10 July 2017. [22] American College of Surgeons. National Cancer Database. https://www.facs.org/quality%20programs/cancer/ncdb; 2017, Accessed date: 10 July 2017. [23] Percy C, Fritz A, Ries L. Conversion of neoplasms by topography and morphology from the International Classification of Diseases for Oncology, Second Edition (ICD-O-2) to the International Classification of Diseases for Oncology, Third Edition (ICD-O-3). https://seer.cancer.gov/tools/ conversion/ICDO2-3manual.pdf; 2001, Accessed date: 15 July 2017. [24] American College of Surgeons. Cancer Program Categories. https:// www.facs.org/quality-programs/cancer/coc/apply/categories, Accessed date: 15 March 2016. [25] National Cancer Institute. NCI-Designated Cancer Centers. http://grants.nih.gov/grants/guide/pa-files/PAR-13-386.html, Accessed date: 15 July 2017. [26] Shuch B, Hofmann JN, Merino MJ, et al. Pathologic validation of renal cell carcinoma histology in the Surveillance, Epidemiology, and End Results program. Urol Oncol 2014;32[23 e29-13]. [27] Lipworth L, Morgans AK, Edwards TL, et al. Renal cell cancer histological subtype distribution differs by race and sex. BJU Int 2016;117:260-5. [28] Olshan AF, Kuo TM, Meyer AM, Nielsen ME, Purdue MP, Rathmell WK. Racial difference in histologic subtype of renal cell carcinoma. Cancer Med 2013;2:744-9.
T. Gansler et al. [29] Purdue MP, Moore LE, Merino MJ, et al. An investigation of risk factors for renal cell carcinoma by histologic subtype in two case-control studies. Int J Cancer 2013;132:2640-7. [30] Griswold ME, Swihart BJ, Caffo BS, Zeger SL. Practical marginalized multilevel models. Stat 2013;2:129-42. [31] Amin MB, Epstein JI, Ulbright TM, et al. Best practices recommendations in the application of immunohistochemistry in urologic pathology: report from the International Society of Urological Pathology consensus conference. Am J Surg Pathol 2014;38:1017-22. [32] Reuter VE, Argani P, Zhou M, Delahunt B, Members of the ISUP Immunohistochemistry in Diagnostic Urologic Pathology Group. Best practices recommendations in the application of immunohistochemistry in the kidney tumors: report from the International Society of Urologic Pathology consensus conference. Am J Surg Pathol 2014;38:e35-9. [33] Cheng L, Amin MB, Beltran-Lopez A, Montironi R. Unclassified renal cell carcinoma. In: Moch H, Humphrey PA, Ulbright TM, Reuter VE, editors. World Health Organization Classification of Tumours. Pathology and Genetics of Tumours of the Urinary System and Male Genital Organs. Lyon, France: IARC Press; 2016. p. 41. [34] Berwick DM. Disseminating innovations in health care. JAMA 2003; 289:1969-75. [35] Gansler T, Fedewa SA, Lin CC, Amin MB, Jemal A, Ward EM. Trends in diagnosis of gleason score 2 through 4 prostate cancer in the National Cancer Database, 1990-2013. Arch Pathol Lab Med 2017; 141:1686-96.