Annual Report to Nation on the Status of Cancer, 1975 to 2011
Trend analysis based on SEER-13 data showed that overall delay-adjusted cancer incidence rates for all persons combined decreased by 0.5% (P < .001) per year from 2002 to 2011 (Table 1). Among men, cancer incidence rates decreased on average by 1.8% (P = .003) annually from 2007 to 2011. Overall cancer incidence rates among women increased 0.8% (P = .003) annually from 1992 to 1998 but were stable from 1998 to 2011. Among children, ages 0–14 and 0–19 years, rates have increased by 0.8% (P < .001) per year over the past decade, continuing a trend dating from 1992.
Among men, delay-adjusted incidence rates from 2002 to 2011 decreased for seven of the most common cancers: prostate (-2.1 AAPC, P < .001), lung and bronchus (lung) (-2.4 AAPC, P < .001), colon and rectum (colorectal) (-3.0 AAPC, P < .001), urinary bladder (bladder) (-0.6 AAPC, P = .05), stomach (-1.7 AAPC, P < .001), brain and other nervous system (brain) (-0.2 AAPC, P = .05), and larynx (-1.9 AAPC, P < .001) (Table 1). Incidence rates among men increased for eight others: melanoma of the skin (melanoma) (2.3 AAPC, P < .001), non-Hodgkin Lymphoma (NHL) (0.3 AAPC, P = .02), kidney and renal pelvis (kidney) (2.0 AAPC, P = .01), leukemia (0.9 AAPC, P = .02), pancreas (1.2 AAPC, P < .001), liver and intrahepatic bile duct (liver) (3.6 AAPC, P < .001), myeloma (1.9 AAPC, P < .001), and thyroid (5.3 AAPC, P < .001). Among women, delay-adjusted incidence rates decreased from 2002 to 2011 for seven of the most common cancers: lung (-1.0 AAPC, P = .001), colorectal (-2.7 AAPC, P < .001), ovary (-0.9 AAPC, P < .001), bladder (-0.9 AAPC, P < .001), cervix uteri (cervix) (-2.4 AAPC, P < .001), oral cavity and pharynx (oral) (-0.7 AAPC, P < .001), and stomach (-0.7 AAPC, P < .001). Incidence rates among women increased for eight others: corpus and uterus (uterus) (1.3 AAPC, P < .001), thyroid (5.8 AAPC, P < .001), melanoma (1.5 AAPC, P < .001), kidney (1.6 AAPC, P = .007), pancreas (1.1 AAPC, P < .001), leukemia (0.6 AAPC, P < .001), myeloma (1.8 AAPC, P = .002), and liver (2.9 AAPC, P < .001). Rates were stable for all other sites, including breast cancer.
Overall cancer death rates have been declining since the early 1990s, with rates from 2002 to 2011 decreasing by about 1.8% (P < .001) per year among males and by 1.4% (P < .001) per year among females (Table 2). Among children ages 0–14 and 0–19 years, rates have continued to decrease since 1975 with a 2.1 AAPC (P < .001) and 2.3 AAPC (P < .001) decrease, respectively, from 2002 to 2011, although decreases were briefly interrupted from 1998 to 2002/2003. During the most recent 10 (2002–2011) and five (2007–2011) data years, death rates among males decreased for 10 top cancers (lung -2.6, P < .001; prostate -3.4, P < .001; colorectal -3.0, P < .001; leukemia -0.9, P < .001; NHL –2.3, P < .001; esophagus -0.5, P < .001; kidney -0.8, P < .001; stomach -3.4, P < .001; myeloma -1.1, P < .001; and larynx -2.5, P < .001 for 2002–2011 AAPC), whereas rates increased from 2002 to 2011 for cancers of the pancreas (0.3 AAPC, P < .001), liver (2.6 AAPC, P < .001), melanoma of the skin (0.3 AAPC, P < .001), and soft tissue including heart (1.1 AAPC, P = .006). During the corresponding time period, death rates among females decreased for 13 of the top cancers (lung -1.2, P < .001; breast -1.9, P < .001; colorectal -2.9, P < .001; ovary -2.0, P < .001; leukemia -1.2, P < .001; NHL -3.2, P < .001; brain -0.9, P < .001; kidney -0.9, P < .001; stomach -2.7, P < .001; cervix -1.3, P < .001; bladder -0.4, P < .001; esophagus -1.5, P < .001; and oral -1.2, P = .004 for 2002–2011 AAPC), whereas they increased from 2002 to 2011 for cancers of the pancreas (0.4 AAPC, P < .001), uterus (1.0 AAPC, P = .001), and liver (1.8 AAPC, P < .001). After decreasing for many years, cancer death rates stabilized between 2007 to 2011 for myeloma among females and for bladder, brain, and oral among males.
Using data submitted to NAACCR from both SEER and NPCR sponsored registries, five-year (2007–2011) average annual incidence rates and five- (2007–2011) and 10-year (2002–2011) incidence trends are shown for the United States (Table 3). During the period between 2007 and 2011, observed rates of all cancers combined in all racial groups were lower among women than for men (412.8 vs 526.1 per 100000). Black men had the highest overall cancer incidence rate (587.7 per 100000) of any racial or ethnic group. Among women, whites had the highest overall cancer incidence rate during this period (418.6 per 100000). Prostate cancer remains the most common cancer among men in each racial and ethnic group and the rates were substantially higher than any other type of cancer. Lung cancer is the second most common cancer and colorectal the third most common cancer among men of all racial and ethnic groups, except in Hispanic men where these ranks reversed. Among women, breast cancer is the most common cancer among all racial and ethnic groups by a wide margin. Lung cancer is also the second most common cancer among women, with colorectal cancer being the third most common cancer, except among API and Hispanic women, where the ranks are again reversed. Rankings of other cancers for both men and women varied by race and ethnicity. White and Hispanic children had higher cancer incidence rates than children of other racial and ethnic groups.
Cancer incidence rates among men declined in each racial/ethnic group, averaging a 1.6% (P < .001) per year decline during the period between 2002 and 2011 with a steeper decline of 2.9% (P = .007) per year during the most recent five years (Table 3). Cancer incidence rates declined among black women and Hispanic women between 2002 and 2011, -0.2 (P ≤ .001) and -0.6 (P = .002) AAPC, respectively, and were stable for women in all other racial/ethnic groups. However, the incidence trend for all women combined during the 2007 to 2011 period showed a decline, averaging 0.9% (P = .04) per year. For children age 0 to 14 and 0 to 19 years, cancer incidence rates increased from 2002 to 2011 for whites (0.5 AAPC, P = .01 and 0.3 AAPC, P = .04, respectively) and non-Hispanic children (0.7 AAPC, P = .002 and 0.5 AAPC, P = .01, respectively), decreased in AI/ANs children (-2.8 AAPC, P = .05 and -2.5 AAPC, P = .01, respectively), and were stable for all other groups.
During the period between 2002 and 2011, the incidence rates for the four most common cancers in men decreased (prostate, lung, colorectal, and bladder) for all races except black and AI/AN men, for whom only prostate, lung, and colorectal cancers declined (Table 3). In addition, stomach (-1.3 AAPC, P < .001), esophageal (-1.1 AAPC, P = .04), brain (-0.7 AAPC, P = .003), and larynx (-2.2 AAPC, P < .001) cancers declined in men for all races combined while kidney, pancreas, liver, and thyroid cancers increased. The trends in males for all races combined were consistent with these findings during the more recent 2007 to 2011 time period, except for kidney cancer, which decreased, and pancreatic and stomach cancer, both of which remained stable. Of particular note was the declining trend for leukemia in the non-delay adjusted data from the NPCR and SEER registries, which directly contrasts with the increasing trend seen in the delay-adjusted SEER data (Table 1).
During the period between 2002 and 2011, lung cancer incidence declined in white, black, and Hispanic women while remaining stable in the other groups (Table 3). Colorectal cancer incidence declined in women in each racial/ethnic group (-3.2 AAPC, P < .001 for all women combined). Overall incidence rates for all women combined declined from 2007 to 2011 (-0.4 AAPC, P = .04) as did ovarian (-2.9 AAPC, P < .001), bladder (-1.2 AAPC, P < .001), cervical (-2.0 AAPC, P < .001), brain (-1.8 AAPC, P < .001), and stomach (-1.1 AAPC, P < .001) cancers. Cancer incidence rates for corpus and uterus (0.9 AAPC, P < .001), thyroid (4.1 AAPC, P < .001), melanoma (1.1 AAPC, P = .03), and liver (2.9 AAPC, P < .001) increased during this time period. On the other hand, breast cancer remained stable among white, AI/AN, and Hispanic women, although slight increases were seen in black and API women. Breast cancer rates were marginally higher in white women compared with black women (124.0 vs 120.7 per 100000 women) and lower in other racial/ethnic groups (Table 3).
For all cancer sites combined, cancer death rates for 2007 through 2011 were higher among men than women (211.6 vs 147.4 deaths per 100000 men) (Table 4). Black men had the highest cancer death rate (269.3 deaths per 100000 men) of any racial or ethnic group. Lung cancer was the leading cause of death in both men and women. Lung, prostate, and colorectal cancers were the leading causes of cancer death among men in every racial and ethnic group except API men, for whom lung, liver, and colorectal ranked highest. For women, the leading causes of cancer death were lung, breast, and colorectal cancers, although the rank order of these top three cancers varied for AI/AN and Hispanic women.
Decreases in overall cancer death rates from 2002 to 2011 were noted for men, women, and children in all racial and ethnic groups, except among API and AI/AN children for whom rates were stable (Table 4). Death rates declined between 2002 and 2011 for the most common cancers (lung, prostate, and colorectal) among men of all racial and ethnic groups except AI/AN. Death rates declined for the top three female cancers (lung, breast, and colorectal) among all racial and ethnic groups; except that rates were stable for lung cancer in API women and for colorectal cancer in AI/AN women. Death rates for liver cancer increased in all subgroups, except for API men, for whom rates decreased, and AI/AN and API women, for whom rates were stable. Pancreatic cancer death rates increased among white men and women. Additionally, death rates for melanoma and soft tissues increased among white men, and death rates from cancers of the uterus increased among white and black women.
A total of 178125 (94.33%) invasive breast cancer cases in states with high quality registries diagnosed in 2011 met our selection criteria (Supplementary Table 1 http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). After imputation, the distribution of HR/HER2 status and associated variables across the original and the imputed datasets looked similar (Supplementary Table 1 http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). The imputed the r value from the model predicting missing HER2 status with available covariates was good (r = 0.39), indicating a good-fitting imputation model. The rates based on the imputed data were higher than the original data because of the imputation-assigned HR/HER2 status, while the general patterns of the age-specific curves looked similar across original and imputed datasets. Figure 1 shows the original and imputed rates for each subtype. The 10 imputations are indistinguishable and overlap. As expected, the imputed rates are higher than the original rates and the magnitude of difference increases with increasing age because the rates of unknown subtype increase with age. For instance, the absolute difference between the original and imputed rate for triple-negative breast cancer for ages 35 to 44 and 75–84 years are 0.2 and 6.4 per 100000, respectively.
(Enlarge Image)
Figure 1.
Original vs imputed age-specific rates by subtype, unknown subtype for diagnosis year 2011, and areas in the United States with high-quality incidence data*^. A) Hormone receptor (HR)+/human epidermal growth factor receptor 2 (HER2)- rates per 100000 women. B) Triple-negative rates per 100000 women. C) HR+HER2+ rates per 100000 women. D) HR-/HER2+ rates per 100000 women. *Population-based registries meeting North American Association of Central Cancer Registries quality criteria and high completeness of HR/HER2 data include: Alaska, California, Colorado, Connecticut, Delaware, District of Columbia, Florida, Georgia, Hawai'i, Idaho, Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Maine, Maryland, Massachusetts, Michigan, Mississippi, Missouri, Montana, Nebraska, New Hampshire, New Jersey, New Mexico, New York, North Carolina, North Dakota, Ohio, Oregon, Pennsylvania, Rhode Island, South Carolina, South Dakota, Utah, Vermont, Virginia, Washington, West Virginia, Wisconsin, Wyoming. ^All 10 imputations had near identical rate estimates. Note: HR+/HER2- has much higher rates, so this figure has a different y-axis. The unknown rate is a reference rate from the original data and is the same for each figure. HER2 = human epidermal growth factor receptor 2; HR = hormone receptor.
Breast cancer subtype HR+/HER2- was the most common subtype, representing 72.6% of all cases, with an age-adjusted rate of 86.5 per 100000; a rate six times higher than triple-negative breast cancer rates of 15.5, seven times higher than HR+/HER2+ breast cancer rate of 12.4, and 16 times higher than HR-/HER2+ breast cancer rate of 5.5 ( Table 5 ). In every race/ethnicity group, rates for HR+/HER2- breast cancers were higher than any other subtype, and HR+/HER2- rate was highest for non-Hispanic white women (92.7 per 100000) (Figure 2; Supplementary Table 2 http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). In women younger than age 45, HR+/HER2- breast cancer rates were comparable among racial/ethnic groups, but for older women rates of this subtype were much higher for non-Hispanic whites than other racial/ethnic groups.
(Enlarge Image)
Figure 2ab
Age-specific incidence rates of invasive breast cancer by subtype, by race/ethnicity, for diagnosis year 2011, and areas in the United States with high-quality incidence data*. A) Age-specific rates for non-Hispanic white women. B) Age-specific rates for non-Hispanic black women.
Rates for triple-negative breast cancers (HR-/HER2-) were highest among non-Hispanic black women compared with all other racial/ethnic groups with an age-adjusted rate of 27.2 per 100000 women; a rate 1.9 times higher than the non-Hispanic white rate, 2.3 times higher than the Hispanic rate, and 2.6 times higher than the non-Hispanic API (NHAPI) rate ( Table 5 ). Triple-negative breast cancers comprised 13% of all breast cancers and were the second most common subtype among non-Hispanic black women in all age groups, after age 45 among non-Hispanic white women, and after age 55 among NHAPI and Hispanic women. Subtype HR-/HER2+ breast cancer (5% of all breast cancers) had the lowest rates for all races/ethnicities, and breast cancer rates of HR+/HER2+ (10% of all breast cancers) were similar to triple-negative rates for all racial/ethnic groups except for non-Hispanic black women, where HR+/HER2+ breast cancer rates were much lower than triple-negative breast cancer rates.
Breast cancers of all subtypes were most commonly diagnosed at a local stage and least commonly diagnosed at a distant stage in all racial/ethnic groups with the highest rate a local stage, 63.51 per 100000, for non-Hispanic white women (Figure 3; Supplementary Table 3 http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). Non-Hispanic black women had the highest rate of breast cancer diagnosed at distant stage across every subtype.
(Enlarge Image)
Figure 2cd
Age-specific incidence rates of invasive breast cancer by subtype, by race/ethnicity, for diagnosis year 2011, and areas in the United States with high-quality incidence data*. C) Age-specific rates for non-Hispanic Asian/Pacific Islander women. D) Age-specific rates for Hispanic women. *Population-based registries meeting North American Association of Central Cancer Registries quality criteria and high completeness of hormone receptor/human epidermal growth factor receptor 2 data include: Alaska, California, Colorado, Connecticut, Delaware, District of Columbia, Florida, Georgia, Hawai'i, Idaho, Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Maine, Maryland, Massachusetts, Michigan, Mississippi, Missouri, Montana, Nebraska, New Hampshire, New Jersey, New Mexico, New York, North Carolina, North Dakota, Ohio, Oregon, Pennsylvania, Rhode Island, South Carolina, South Dakota, Utah, Vermont, Virginia, Washington, West Virginia, Wisconsin, Wyoming. HER2 = human epidermal growth factor receptor 2; HR = hormone receptor.
Differences in tumor grade were observed across breast cancer subtypes. Among HR+/HER2- breast cancer cases, rates of moderately differentiated breast cancer were highest for all racial/ethnic groups, and rates of the least favorable grades, poorly differentiated and undifferentiated, were lowest for all groups except for non-Hispanic black women (Figure 4; Supplementary Table 4 http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). For all other breast cancer subtypes, rates of poorly/undifferentiated grade cases greatly exceeded the more favorable grades in every racial/ethnic group. Rates for poorly and undifferentiated cases were highest for triple-negative breast cancers among non-Hispanic black women.
Breast cancer rates of HR+/HER2- decreased with increasing poverty for every racial and ethnic group with the highest rate, 98.69 per 100000, for non-Hispanic white women living in low poverty areas (Figure 5; Supplementary Table 5 http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). There were no clear relationships between census tract-based poverty and incidence for the other subtypes for any race/ethnicity.
The geographic distribution of breast cancer by subtype is shown in Figure 6. Because of small cell size, we were unable to stratify our state-level analysis by race/ethnicity. States with rates that were statistically higher or lower than the overall national rate are identifiable through the bar graphs to the left of the maps. State-level triple-negative breast cancers rates were lower in the northwest and higher in the southeast (Figure 6). Rates of HR+/HER2+ breast cancer were higher than the national rate in Idaho, Tennessee, and Pennsylvania and statistically lower in Colorado, Florida, Hawai'i, Kentucky, Maine, South Dakota, and Virginia. For HR-/HER2+ breast cancer, no states had rates that were statistically different from the national rate.
These maps were descriptive, ecologic assessments of the data. Geographic variation is driven by multiple individual and system-level factors, and the state-level differences must be interpreted with prudence. With this in mind, incidence rates of HR+/HER2- breast cancers were generally higher in states with higher mammography screening rates (Supplementary Figure 1A http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). Correlation analysis indicated HR+/HER2- breast cancer rates were highly correlated with self-reported mammography rates for non-Hispanic white women (Pearson r = 0.57, P < .001; Spearman ρ = 0.58, P < .001) and moderately correlated for non-Hispanic blacks, non-Hispanic API, and Hispanic women combined (Pearson r = .33, P = .033; Spearman ρ = 0.32, P = .037). Triple-negative cancers decreased with increasing percent of mammography for non-Hispanic Asian and Pacific Islanders (Pearson r = -0.46, P = 0.19; Spearman ρ = -0.45, P = .021), however, the cell counts in many states were too small to be stable. No correlations with mammography were identified for the other subtypes. Triple-negative breast cancer rates increased with increasing percent of non-Hispanic black population (Supplementary Figure 1B http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online), and the association was strongly correlated (Pearson r = 0.80, P < .001; Spearman ρ = 0.73, P < .001). No correlations with race/ethnicity were identified for other subtypes.
Results
Cancer Incidence Rate Long-term Trends (1992-2011) for Most Common Cancers
Trend analysis based on SEER-13 data showed that overall delay-adjusted cancer incidence rates for all persons combined decreased by 0.5% (P < .001) per year from 2002 to 2011 (Table 1). Among men, cancer incidence rates decreased on average by 1.8% (P = .003) annually from 2007 to 2011. Overall cancer incidence rates among women increased 0.8% (P = .003) annually from 1992 to 1998 but were stable from 1998 to 2011. Among children, ages 0–14 and 0–19 years, rates have increased by 0.8% (P < .001) per year over the past decade, continuing a trend dating from 1992.
Among men, delay-adjusted incidence rates from 2002 to 2011 decreased for seven of the most common cancers: prostate (-2.1 AAPC, P < .001), lung and bronchus (lung) (-2.4 AAPC, P < .001), colon and rectum (colorectal) (-3.0 AAPC, P < .001), urinary bladder (bladder) (-0.6 AAPC, P = .05), stomach (-1.7 AAPC, P < .001), brain and other nervous system (brain) (-0.2 AAPC, P = .05), and larynx (-1.9 AAPC, P < .001) (Table 1). Incidence rates among men increased for eight others: melanoma of the skin (melanoma) (2.3 AAPC, P < .001), non-Hodgkin Lymphoma (NHL) (0.3 AAPC, P = .02), kidney and renal pelvis (kidney) (2.0 AAPC, P = .01), leukemia (0.9 AAPC, P = .02), pancreas (1.2 AAPC, P < .001), liver and intrahepatic bile duct (liver) (3.6 AAPC, P < .001), myeloma (1.9 AAPC, P < .001), and thyroid (5.3 AAPC, P < .001). Among women, delay-adjusted incidence rates decreased from 2002 to 2011 for seven of the most common cancers: lung (-1.0 AAPC, P = .001), colorectal (-2.7 AAPC, P < .001), ovary (-0.9 AAPC, P < .001), bladder (-0.9 AAPC, P < .001), cervix uteri (cervix) (-2.4 AAPC, P < .001), oral cavity and pharynx (oral) (-0.7 AAPC, P < .001), and stomach (-0.7 AAPC, P < .001). Incidence rates among women increased for eight others: corpus and uterus (uterus) (1.3 AAPC, P < .001), thyroid (5.8 AAPC, P < .001), melanoma (1.5 AAPC, P < .001), kidney (1.6 AAPC, P = .007), pancreas (1.1 AAPC, P < .001), leukemia (0.6 AAPC, P < .001), myeloma (1.8 AAPC, P = .002), and liver (2.9 AAPC, P < .001). Rates were stable for all other sites, including breast cancer.
Long-term (1975–2011) Cancer Mortality Trends for all Racial and Ethnic Groups Combined
Overall cancer death rates have been declining since the early 1990s, with rates from 2002 to 2011 decreasing by about 1.8% (P < .001) per year among males and by 1.4% (P < .001) per year among females (Table 2). Among children ages 0–14 and 0–19 years, rates have continued to decrease since 1975 with a 2.1 AAPC (P < .001) and 2.3 AAPC (P < .001) decrease, respectively, from 2002 to 2011, although decreases were briefly interrupted from 1998 to 2002/2003. During the most recent 10 (2002–2011) and five (2007–2011) data years, death rates among males decreased for 10 top cancers (lung -2.6, P < .001; prostate -3.4, P < .001; colorectal -3.0, P < .001; leukemia -0.9, P < .001; NHL –2.3, P < .001; esophagus -0.5, P < .001; kidney -0.8, P < .001; stomach -3.4, P < .001; myeloma -1.1, P < .001; and larynx -2.5, P < .001 for 2002–2011 AAPC), whereas rates increased from 2002 to 2011 for cancers of the pancreas (0.3 AAPC, P < .001), liver (2.6 AAPC, P < .001), melanoma of the skin (0.3 AAPC, P < .001), and soft tissue including heart (1.1 AAPC, P = .006). During the corresponding time period, death rates among females decreased for 13 of the top cancers (lung -1.2, P < .001; breast -1.9, P < .001; colorectal -2.9, P < .001; ovary -2.0, P < .001; leukemia -1.2, P < .001; NHL -3.2, P < .001; brain -0.9, P < .001; kidney -0.9, P < .001; stomach -2.7, P < .001; cervix -1.3, P < .001; bladder -0.4, P < .001; esophagus -1.5, P < .001; and oral -1.2, P = .004 for 2002–2011 AAPC), whereas they increased from 2002 to 2011 for cancers of the pancreas (0.4 AAPC, P < .001), uterus (1.0 AAPC, P = .001), and liver (1.8 AAPC, P < .001). After decreasing for many years, cancer death rates stabilized between 2007 to 2011 for myeloma among females and for bladder, brain, and oral among males.
Cancer Incidence Rates (2007-2011) and Trends (2007-2011 and 2002-2011) by Race and Ethnicity
Using data submitted to NAACCR from both SEER and NPCR sponsored registries, five-year (2007–2011) average annual incidence rates and five- (2007–2011) and 10-year (2002–2011) incidence trends are shown for the United States (Table 3). During the period between 2007 and 2011, observed rates of all cancers combined in all racial groups were lower among women than for men (412.8 vs 526.1 per 100000). Black men had the highest overall cancer incidence rate (587.7 per 100000) of any racial or ethnic group. Among women, whites had the highest overall cancer incidence rate during this period (418.6 per 100000). Prostate cancer remains the most common cancer among men in each racial and ethnic group and the rates were substantially higher than any other type of cancer. Lung cancer is the second most common cancer and colorectal the third most common cancer among men of all racial and ethnic groups, except in Hispanic men where these ranks reversed. Among women, breast cancer is the most common cancer among all racial and ethnic groups by a wide margin. Lung cancer is also the second most common cancer among women, with colorectal cancer being the third most common cancer, except among API and Hispanic women, where the ranks are again reversed. Rankings of other cancers for both men and women varied by race and ethnicity. White and Hispanic children had higher cancer incidence rates than children of other racial and ethnic groups.
Cancer incidence rates among men declined in each racial/ethnic group, averaging a 1.6% (P < .001) per year decline during the period between 2002 and 2011 with a steeper decline of 2.9% (P = .007) per year during the most recent five years (Table 3). Cancer incidence rates declined among black women and Hispanic women between 2002 and 2011, -0.2 (P ≤ .001) and -0.6 (P = .002) AAPC, respectively, and were stable for women in all other racial/ethnic groups. However, the incidence trend for all women combined during the 2007 to 2011 period showed a decline, averaging 0.9% (P = .04) per year. For children age 0 to 14 and 0 to 19 years, cancer incidence rates increased from 2002 to 2011 for whites (0.5 AAPC, P = .01 and 0.3 AAPC, P = .04, respectively) and non-Hispanic children (0.7 AAPC, P = .002 and 0.5 AAPC, P = .01, respectively), decreased in AI/ANs children (-2.8 AAPC, P = .05 and -2.5 AAPC, P = .01, respectively), and were stable for all other groups.
During the period between 2002 and 2011, the incidence rates for the four most common cancers in men decreased (prostate, lung, colorectal, and bladder) for all races except black and AI/AN men, for whom only prostate, lung, and colorectal cancers declined (Table 3). In addition, stomach (-1.3 AAPC, P < .001), esophageal (-1.1 AAPC, P = .04), brain (-0.7 AAPC, P = .003), and larynx (-2.2 AAPC, P < .001) cancers declined in men for all races combined while kidney, pancreas, liver, and thyroid cancers increased. The trends in males for all races combined were consistent with these findings during the more recent 2007 to 2011 time period, except for kidney cancer, which decreased, and pancreatic and stomach cancer, both of which remained stable. Of particular note was the declining trend for leukemia in the non-delay adjusted data from the NPCR and SEER registries, which directly contrasts with the increasing trend seen in the delay-adjusted SEER data (Table 1).
During the period between 2002 and 2011, lung cancer incidence declined in white, black, and Hispanic women while remaining stable in the other groups (Table 3). Colorectal cancer incidence declined in women in each racial/ethnic group (-3.2 AAPC, P < .001 for all women combined). Overall incidence rates for all women combined declined from 2007 to 2011 (-0.4 AAPC, P = .04) as did ovarian (-2.9 AAPC, P < .001), bladder (-1.2 AAPC, P < .001), cervical (-2.0 AAPC, P < .001), brain (-1.8 AAPC, P < .001), and stomach (-1.1 AAPC, P < .001) cancers. Cancer incidence rates for corpus and uterus (0.9 AAPC, P < .001), thyroid (4.1 AAPC, P < .001), melanoma (1.1 AAPC, P = .03), and liver (2.9 AAPC, P < .001) increased during this time period. On the other hand, breast cancer remained stable among white, AI/AN, and Hispanic women, although slight increases were seen in black and API women. Breast cancer rates were marginally higher in white women compared with black women (124.0 vs 120.7 per 100000 women) and lower in other racial/ethnic groups (Table 3).
Current Cancer Death Rates (2007–2011) and Trends (2002–2011 and 2007–2011) by Race and Ethnicity
For all cancer sites combined, cancer death rates for 2007 through 2011 were higher among men than women (211.6 vs 147.4 deaths per 100000 men) (Table 4). Black men had the highest cancer death rate (269.3 deaths per 100000 men) of any racial or ethnic group. Lung cancer was the leading cause of death in both men and women. Lung, prostate, and colorectal cancers were the leading causes of cancer death among men in every racial and ethnic group except API men, for whom lung, liver, and colorectal ranked highest. For women, the leading causes of cancer death were lung, breast, and colorectal cancers, although the rank order of these top three cancers varied for AI/AN and Hispanic women.
Decreases in overall cancer death rates from 2002 to 2011 were noted for men, women, and children in all racial and ethnic groups, except among API and AI/AN children for whom rates were stable (Table 4). Death rates declined between 2002 and 2011 for the most common cancers (lung, prostate, and colorectal) among men of all racial and ethnic groups except AI/AN. Death rates declined for the top three female cancers (lung, breast, and colorectal) among all racial and ethnic groups; except that rates were stable for lung cancer in API women and for colorectal cancer in AI/AN women. Death rates for liver cancer increased in all subgroups, except for API men, for whom rates decreased, and AI/AN and API women, for whom rates were stable. Pancreatic cancer death rates increased among white men and women. Additionally, death rates for melanoma and soft tissues increased among white men, and death rates from cancers of the uterus increased among white and black women.
HR/HER2 Breast Cancer Subtypes
A total of 178125 (94.33%) invasive breast cancer cases in states with high quality registries diagnosed in 2011 met our selection criteria (Supplementary Table 1 http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). After imputation, the distribution of HR/HER2 status and associated variables across the original and the imputed datasets looked similar (Supplementary Table 1 http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). The imputed the r value from the model predicting missing HER2 status with available covariates was good (r = 0.39), indicating a good-fitting imputation model. The rates based on the imputed data were higher than the original data because of the imputation-assigned HR/HER2 status, while the general patterns of the age-specific curves looked similar across original and imputed datasets. Figure 1 shows the original and imputed rates for each subtype. The 10 imputations are indistinguishable and overlap. As expected, the imputed rates are higher than the original rates and the magnitude of difference increases with increasing age because the rates of unknown subtype increase with age. For instance, the absolute difference between the original and imputed rate for triple-negative breast cancer for ages 35 to 44 and 75–84 years are 0.2 and 6.4 per 100000, respectively.
(Enlarge Image)
Figure 1.
Original vs imputed age-specific rates by subtype, unknown subtype for diagnosis year 2011, and areas in the United States with high-quality incidence data*^. A) Hormone receptor (HR)+/human epidermal growth factor receptor 2 (HER2)- rates per 100000 women. B) Triple-negative rates per 100000 women. C) HR+HER2+ rates per 100000 women. D) HR-/HER2+ rates per 100000 women. *Population-based registries meeting North American Association of Central Cancer Registries quality criteria and high completeness of HR/HER2 data include: Alaska, California, Colorado, Connecticut, Delaware, District of Columbia, Florida, Georgia, Hawai'i, Idaho, Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Maine, Maryland, Massachusetts, Michigan, Mississippi, Missouri, Montana, Nebraska, New Hampshire, New Jersey, New Mexico, New York, North Carolina, North Dakota, Ohio, Oregon, Pennsylvania, Rhode Island, South Carolina, South Dakota, Utah, Vermont, Virginia, Washington, West Virginia, Wisconsin, Wyoming. ^All 10 imputations had near identical rate estimates. Note: HR+/HER2- has much higher rates, so this figure has a different y-axis. The unknown rate is a reference rate from the original data and is the same for each figure. HER2 = human epidermal growth factor receptor 2; HR = hormone receptor.
Breast cancer subtype HR+/HER2- was the most common subtype, representing 72.6% of all cases, with an age-adjusted rate of 86.5 per 100000; a rate six times higher than triple-negative breast cancer rates of 15.5, seven times higher than HR+/HER2+ breast cancer rate of 12.4, and 16 times higher than HR-/HER2+ breast cancer rate of 5.5 ( Table 5 ). In every race/ethnicity group, rates for HR+/HER2- breast cancers were higher than any other subtype, and HR+/HER2- rate was highest for non-Hispanic white women (92.7 per 100000) (Figure 2; Supplementary Table 2 http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). In women younger than age 45, HR+/HER2- breast cancer rates were comparable among racial/ethnic groups, but for older women rates of this subtype were much higher for non-Hispanic whites than other racial/ethnic groups.
(Enlarge Image)
Figure 2ab
Age-specific incidence rates of invasive breast cancer by subtype, by race/ethnicity, for diagnosis year 2011, and areas in the United States with high-quality incidence data*. A) Age-specific rates for non-Hispanic white women. B) Age-specific rates for non-Hispanic black women.
Rates for triple-negative breast cancers (HR-/HER2-) were highest among non-Hispanic black women compared with all other racial/ethnic groups with an age-adjusted rate of 27.2 per 100000 women; a rate 1.9 times higher than the non-Hispanic white rate, 2.3 times higher than the Hispanic rate, and 2.6 times higher than the non-Hispanic API (NHAPI) rate ( Table 5 ). Triple-negative breast cancers comprised 13% of all breast cancers and were the second most common subtype among non-Hispanic black women in all age groups, after age 45 among non-Hispanic white women, and after age 55 among NHAPI and Hispanic women. Subtype HR-/HER2+ breast cancer (5% of all breast cancers) had the lowest rates for all races/ethnicities, and breast cancer rates of HR+/HER2+ (10% of all breast cancers) were similar to triple-negative rates for all racial/ethnic groups except for non-Hispanic black women, where HR+/HER2+ breast cancer rates were much lower than triple-negative breast cancer rates.
Breast cancers of all subtypes were most commonly diagnosed at a local stage and least commonly diagnosed at a distant stage in all racial/ethnic groups with the highest rate a local stage, 63.51 per 100000, for non-Hispanic white women (Figure 3; Supplementary Table 3 http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). Non-Hispanic black women had the highest rate of breast cancer diagnosed at distant stage across every subtype.
(Enlarge Image)
Figure 2cd
Age-specific incidence rates of invasive breast cancer by subtype, by race/ethnicity, for diagnosis year 2011, and areas in the United States with high-quality incidence data*. C) Age-specific rates for non-Hispanic Asian/Pacific Islander women. D) Age-specific rates for Hispanic women. *Population-based registries meeting North American Association of Central Cancer Registries quality criteria and high completeness of hormone receptor/human epidermal growth factor receptor 2 data include: Alaska, California, Colorado, Connecticut, Delaware, District of Columbia, Florida, Georgia, Hawai'i, Idaho, Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Maine, Maryland, Massachusetts, Michigan, Mississippi, Missouri, Montana, Nebraska, New Hampshire, New Jersey, New Mexico, New York, North Carolina, North Dakota, Ohio, Oregon, Pennsylvania, Rhode Island, South Carolina, South Dakota, Utah, Vermont, Virginia, Washington, West Virginia, Wisconsin, Wyoming. HER2 = human epidermal growth factor receptor 2; HR = hormone receptor.
Differences in tumor grade were observed across breast cancer subtypes. Among HR+/HER2- breast cancer cases, rates of moderately differentiated breast cancer were highest for all racial/ethnic groups, and rates of the least favorable grades, poorly differentiated and undifferentiated, were lowest for all groups except for non-Hispanic black women (Figure 4; Supplementary Table 4 http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). For all other breast cancer subtypes, rates of poorly/undifferentiated grade cases greatly exceeded the more favorable grades in every racial/ethnic group. Rates for poorly and undifferentiated cases were highest for triple-negative breast cancers among non-Hispanic black women.
Breast cancer rates of HR+/HER2- decreased with increasing poverty for every racial and ethnic group with the highest rate, 98.69 per 100000, for non-Hispanic white women living in low poverty areas (Figure 5; Supplementary Table 5 http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). There were no clear relationships between census tract-based poverty and incidence for the other subtypes for any race/ethnicity.
The geographic distribution of breast cancer by subtype is shown in Figure 6. Because of small cell size, we were unable to stratify our state-level analysis by race/ethnicity. States with rates that were statistically higher or lower than the overall national rate are identifiable through the bar graphs to the left of the maps. State-level triple-negative breast cancers rates were lower in the northwest and higher in the southeast (Figure 6). Rates of HR+/HER2+ breast cancer were higher than the national rate in Idaho, Tennessee, and Pennsylvania and statistically lower in Colorado, Florida, Hawai'i, Kentucky, Maine, South Dakota, and Virginia. For HR-/HER2+ breast cancer, no states had rates that were statistically different from the national rate.
These maps were descriptive, ecologic assessments of the data. Geographic variation is driven by multiple individual and system-level factors, and the state-level differences must be interpreted with prudence. With this in mind, incidence rates of HR+/HER2- breast cancers were generally higher in states with higher mammography screening rates (Supplementary Figure 1A http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online). Correlation analysis indicated HR+/HER2- breast cancer rates were highly correlated with self-reported mammography rates for non-Hispanic white women (Pearson r = 0.57, P < .001; Spearman ρ = 0.58, P < .001) and moderately correlated for non-Hispanic blacks, non-Hispanic API, and Hispanic women combined (Pearson r = .33, P = .033; Spearman ρ = 0.32, P = .037). Triple-negative cancers decreased with increasing percent of mammography for non-Hispanic Asian and Pacific Islanders (Pearson r = -0.46, P = 0.19; Spearman ρ = -0.45, P = .021), however, the cell counts in many states were too small to be stable. No correlations with mammography were identified for the other subtypes. Triple-negative breast cancer rates increased with increasing percent of non-Hispanic black population (Supplementary Figure 1B http://jnci.oxfordjournals.org/content/107/6/djv048/suppl/DC1, available online), and the association was strongly correlated (Pearson r = 0.80, P < .001; Spearman ρ = 0.73, P < .001). No correlations with race/ethnicity were identified for other subtypes.