Breast cancer is rising fastest in Korea and Japan — and slowest in Asian Americans

The standard summary of global breast cancer epidemiology is that incidence is high in Western countries and low in East Asia. An age-period-cohort analysis published online on 1 May shows that the second half of that sentence is becoming obsolete, and does something more useful than noting the convergence: it compares native Asian populations against their migrant counterparts in the United States, which separates what is changing in Asia from what changes when people move [s1].

The data

The authors analysed population-based cancer registry data from Japan (1985–2019), Korea (1993–2017), Hawaii (1988–2017), Los Angeles County (1988–2017) and the SEER 8 registries (1985–2019), covering women aged 25 to 84 grouped into five-year age bands and calendar periods [s1]. Age-period-cohort analyses were run using the National Cancer Institute's APC web-based tool [s1].

Six populations were compared: native Japanese, Japanese American, native Korean, Korean American, US White and US Black women [s1].

The rates of change

The annual increases separate cleanly by geography:

• Native Korean women: 5.3% [s1]
• Native Japanese women: 3.8% [s1]
• Korean American women: 2.8% [s1]
• Japanese American women: 1.6% [s1]
• US White women: stable or declining [s1]
• US Black women: modest increases [s1]

The ordering is the finding. Each Asian-origin population in the United States is rising more slowly than its counterpart population in Asia, and both are rising faster than US White women, whose rates are flat or falling.

The shape of the curves, not just their slope

Incidence rose with age in every population, but the patterns differed [s1]. Native Japanese and Korean women had the lowest overall rates, with an early-midlife peak — incidence rising to a maximum in the middle years rather than continuing to climb [s1]. Japanese American women resembled US White women, with incidence increasing into older ages [s1].

That difference in curve shape matters more than it sounds. A population with an early-midlife peak and a population with rates climbing into the 70s and 80s have different age distributions of cases, which implies different screening ages, different absolute case counts at a given population age structure, and different treatment demand — even at similar crude rates.

Period and cohort effects were strongest among native Japanese and Korean women, with sustained increases among women born after 1950 [s1].

What a birth-cohort effect implies

Age-period-cohort analysis separates three things that ordinary trend data confounds: risk that varies with age, risk that shifts for everyone at once in a given calendar period, and risk carried by a birth cohort through its life.

A strong cohort effect starting with women born after 1950 points at exposures that differ between generations rather than at something that happened in a particular calendar year. What those exposures are is not established here. This is a registry analysis of incidence, not an aetiological study: it measures when and to whom cases occurred, and does not measure any risk factor.

The same caution applies in the other direction. Period effects in cancer registry data can reflect detection as well as disease — anything that increases the rate at which cancers are found will raise measured incidence independently of underlying risk. A registry-based APC analysis cannot fully separate the two. A cohort effect sustained across successive generations is harder to explain by detection alone than a period effect is, but it is not immune to it either.

Why the migrant comparison is the useful part

The classic migrant-study logic runs: if risk changes when a population moves, environment and behaviour dominate; if it does not, inherited risk dominates. This analysis runs that comparison directly, and the answer has two halves that point in different directions.

In the shape of the age curve, the migrant populations look American: Japanese American women resembled US White women, with incidence increasing into older ages, while native Japanese and Korean women peaked in early midlife [s1]. But in the rate of change, the migrant populations are the slower ones — native Koreans at 5.3% and native Japanese at 3.8% a year against 2.8% for Korean Americans and 1.6% for Japanese Americans [s1]. Whatever convergence is happening between East Asian and US incidence, the faster movement is on the Asian side.

The authors describe this as an ongoing epidemiologic transition in East Asia, and as a demonstration of how migration and generational change shape breast cancer risk across populations [s1].

Limits

Registry coverage periods differ across the six populations — Korea's data starts in 1993, Japan's and SEER's in 1985 — so the populations are not compared over identical windows [s1]. Registries also differ in completeness and in how race and ethnicity are recorded, and Hawaii and Los Angeles County are two specific places rather than the whole Japanese American or Korean American population. None of that undermines the direction of the finding, but it argues against reading the percentage differences as precise.

What to watch

Whether the native Korean and Japanese increases continue at these rates through more recent registry years, and whether the early-midlife peak in native populations persists as the post-1950 cohorts age into their 70s. If it does not — if those cohorts carry their elevated risk into older ages the way US populations do — the case burden implied is considerably larger than current rates suggest.

This article is informational and does not constitute medical advice.