Page 59 - An Evidence Review of Active Surveillance in Men With Localized Prostate Cancer

prostate cancer death (more than two-fold and more than four-fold higher, respectively). These
differences were relatively constant over the time period covered by the study (deaths occurring
24
during 1988-95). These findings were confirmed by the third study which, using data from the
SEER-Detroit database, showed that patients with grade I tumors have the highest 5-year relative
survival compared to those with tumors of higher grade and that improvements in this statistic
have occurred for all tumor grades (1 to 4) during the study period (patients diagnosed up to
86
1997 and followed up to 2001). Trends in mortality rates stratified by tumor grade should be
interpreted cautiously given temporal changes in histopathologic grading practice (see next
section, subsection on “Histopathologic grading changes”).
We did not identify any studies reporting on population trends in mortality rates stratified by
tumor grade.

Patient, Tumor, and System-Level Characteristics at Diagnosis

Patient Characteristics
We identified 52 observational studies reporting on patient characteristics at presentation (27
SEER, 7 SEER-Medicare, 10 CaPSURE, 5 NCDB and 3 other databases) (Appendix Tables
C1.4-1.14).
The most commonly examined characteristics at baseline were patient age (21
studies), 19,21,29,30,37,42,46,50,52,55,56,63,65,71-75,77,86,91 race/ethnicity (18 studies), 31,37,46,49,52,53,58,61-
63,65,71,74,75,77,80,86,92 tumor grade (16 studies), 4,22,30,37,46,52,56,65,67,68,71,73,75,77,86,90 and tumor stage (22
studies). 4,16,23,31,37,46,52,56,62,63,65,67,68,71-76,86,90,91 Information was available for all factors relevant to
Key Question 1 except tumor volume. Studies were generally large (median sample = 46,248;
th
th
25 -75 percentile, 10,385-134,434), published between 1990 and 2011, and covered the years
1973 to 2008.

Age. Twenty one studies (12 SEER or its component registries, 1 SEER-Medicare, 2 CaPSURE,
5 NCDB, 1 Patterns of Care Study (POCS) data) covering 1973 to 2005, reported information
regarding patients’ age at presentation (Appendix Table C1.4). 19,21,29,30,37,42,46,50,52,55,56,63,65,71-
75,77,86,91
Seven studies (5 SEER or its component registries, 2 NCDB) reported trends in the average
(mean or median) age at diagnosis of prostate cancer. Five of these studies reported reductions in
the average age of patients whereas two studies did not report any changes during their
respective time periods. Only one of these studies covered the period after 2005: using the SEER
database, this study reported a statistically significant reduction over time in the mean age at
42
diagnosis (from 72.2 to 67.2 years, comparing 1988-89 versus 2004-05). This change was
statistically significant (P < 0.001) and was observed both for whites (absolute reduction = 4.7
years) and blacks (absolute reduction = 6.4 years). Notably, the three studies (all using the SEER
database) that reported information on trends in average age stratified by race found that blacks
were diagnosed at a younger average age than whites and that this difference persisted over time
(i.e., despite changes in the race-specific average age at diagnosis); 19,21,42 no such analyses were
reported for individuals belonging to other racial/ethnic groups.
Fifteen studies (7 SEER or its component registries, 1 SEER-Medicare, 2 CaPSURE, 4
NCDB, and 1 POCS) reported the distribution of patients across discrete age groups (one study
using the NCDB database provided information on age both as a continuous variable and
grouped into discrete categories and is also included in the preceding paragraph) and generally

24