Unpacking "Cancer Statistics, 2025": Hidden Crises in Survival and Stage Distribution
A review of the American Cancer Society's 2025 cancer statistics, including population-level trends, stage at diagnosis, and important limits on interpreting registry estimates.
1. The Gold Standard of Oncological Data: SEER and NCHS
Every year, the oncology community highly anticipates the publication of the American Cancer Society's "Cancer Statistics" report[1]. The immense credibility of this annual publication stems from its robust, government-backed data sources. Population-based cancer incidence data are meticulously collected by the National Cancer Institute's (NCI) Surveillance, Epidemiology, and End Results (SEER) program, alongside the Centers for Disease Control and Prevention's (CDC) National Program of Cancer Registries (NPCR)[1]. Mortality data are compiled by the National Center for Health Statistics (NCHS)[1].
Because the combined incidence data from registries participating in SEER and the CDC's National Program of Cancer Registries approach nearly 100% coverage of the US population for recent years[1], they provide a broad population-level view of cancer trends. However, descriptive registry data alone cannot determine how much of a trend was caused by a particular screening program, treatment, or other factor.
2. The 2025 Projection: A Shifting Demographic Burden
For 2025, the report projects 2,041,910 new invasive cancer cases and 618,120 cancer deaths in the United States[1]. While the overall cancer mortality rate has continued to decline through 2022—averting nearly 4.5 million deaths since 1991 [1]—the incidence data reveal a concerning demographic shift.
Historically, cancer incidence has been higher in men due to greater exposure to carcinogenic lifestyle factors like smoking[1]. However, overall cancer incidence has generally declined in men but has steadily risen in women[1]. The gap is closing rapidly: incidence rates in women aged 50-64 have already surpassed those in men (832.5 vs. 830.6 per 100,000)[1]. Even more striking, women younger than 50 years now have an 82% higher incidence rate than their male counterparts, up from 51% in 2002[1]. Notably, in 2021, lung cancer incidence in women surpassed that in men among people younger than 65 years for the first time[1].
3. Ovarian Cancer: Interpreting Survival Trends Alongside Stage at Diagnosis
When analyzing the efficacy of cancer treatments, 5-year relative survival rates are often the primary metric. However, looking at survival rates in isolation can be dangerously misleading. A prime example highlighted in the SEER data is Ovarian Cancer.
According to the historic data, the 5-year relative survival rate for ovarian cancer has steadily improved from 36% (1975-1977) to 43% (1995-1997), reaching 51% in the most recent cohort (2014-2020)[1]. On paper, a jump from 36% to 51% seems like a solid therapeutic victory. Yet, in 2025, an estimated 20,890 women will be diagnosed with ovarian cancer, and a staggering 12,730 will die from it[1]. Why is the death toll so high if survival rates are improving?
Stage distribution is one important part of the explanation: the report shows that many ovarian cancers are diagnosed at a distant stage rather than while localized or regional[1]. Later stage at diagnosis is associated with worse outcomes, but these descriptive statistics do not prove that a proposed screening test would reduce mortality. Ovarian cancer does not currently have a recommended population-wide screening strategy for average-risk people; any proposed early-detection biomarker would need evidence of acceptable test performance and improved clinical outcomes, not only earlier diagnosis.
4. A Half-Century of Triumphs and the Lone Anomaly
The 50-year comparison (1975 to 2020) shows substantial improvements in reported 5-year relative survival: from 49% to 69% for all cancers combined[1], from 34% to 67% for leukemia, and from 68% to 97% for prostate cancer[1]. These are important population-level trends, but they should not be attributed to individual therapies or screening practices on the basis of survival comparisons alone. Changes in incidence, stage at diagnosis, classification, treatment, and other factors may all contribute; for prostate cancer in particular, PSA screening can also affect measured survival through earlier diagnosis and lead-time effects.
The report also highlights areas of concern. Uterine corpus (endometrial) cancer is the only cancer listed in this comparison for which 5-year relative survival decreased, from 87% (1975–1977) to 81% (2014–2020)[1]. The report discusses changes in incidence and histologic subtype as possible contributors; this population-level comparison cannot quantify the contribution of each factor or establish a single cause. Differences in access to care and outcomes across populations also require careful investigation rather than being treated as a complete causal explanation from registry data alone[1].
5. Persistent and Alarming Disparities
Population-based registries can reveal important disparities in cancer outcomes. In the report, some American Indian and Alaska Native populations have substantially higher mortality rates than White populations for selected cancers, while Black populations have higher mortality rates for several cancer types, including prostate, stomach, and uterine corpus cancers[1]. These patterns vary by cancer site, geography, and population definition; registry comparisons describe disparities but do not, on their own, identify the relative contribution of access, exposure, tumor biology, or other causes.
Toolkit Tip: When working with epidemiological registry data like SEER, comparing expected versus observed survival frequencies across different demographics (e.g., Localized vs. Distant stage diagnosis by race) is critical. Use our Chi-Square Calculator to determine if the stage distribution discrepancies in your own institutional cohorts are statistically significant.
How to Read Cancer Statistics Without Overinterpreting Them
- Counts are not rates. The number of diagnoses or deaths can rise as the population grows or ages even when an age-adjusted rate is stable or falling.
- Incidence and mortality answer different questions. Incidence reflects new diagnoses; mortality reflects deaths. Screening, diagnostic practice, treatment, and follow-up can affect these measures differently.
- Age adjustment enables fairer comparisons, not individual predictions. Standardized rates help compare populations with different age structures but do not describe any one person's absolute risk.
- Check geography and period. U.S. registry-based estimates should not be transferred directly to Korea or another country, where population structure, screening, exposures, and health systems differ.
- Read uncertainty and data source notes. Recent-year estimates may rely on projections or modeled values when complete registry data are not yet available. Verify the original report's definitions before quoting a number.
These distinctions matter when communicating trends to patients, policy teams, or investors: a single headline statistic rarely captures the whole disease burden.