Abstract P5-07-07: A generalizable relationship between mortality and time-to-death among breast cancer patients can be explained by tumor dormancy
Bibliographic record
Abstract
Abstract Background: Women with ER-positive breast cancer may recur as late as twenty years post diagnosis. The reason for this delayed recurrence is unknown. We studied survival patterns, including time-to-death in 123,705 women with stage I to III invasive breast cancer, enrolled in the SEER database. We sought to determine if time from diagnosis to death could be explained by variation in tumor dormancy. Methods: We divided the cohort into ten classes based on their 20-year risks of death from breast cancer. Probability of death at 20 years was estimated for each patient using year of diagnosis, patient characteristics (age, race, household income), tumor characteristics (grade, tumor size, nodal status, stage, ER and PR-status) and treatment (radiotherapy, chemotherapy). The 20-year mortality for women in the highest risk decile (decile 10) was 69% versus 5% for women in the lowest decile (decile 1). We define a variable α which represents the annual rate of reactivation from tumor dormancy. Using the highest risk decile as a reference distribution, we determined what α value gives the best fit for the time-to-death distribution of each decile. As a complementary analysis we performed quantile regression to determine independent predictors of time-to-death among women who died from breast cancer. Results: The duration of tumor dormancy was much longer, on average, for ER-positive breast cancers than for ER-negative breast cancers. Time at peak mortality rate monotonically decreased by decile, ranging from 19 years post-diagnosis in the lowest risk decile to two years post-diagnosis in the highest risk decile (Table 1). The difference in the time-to-death distributions by decile could be explained by the variable α which represents the annual rate of reactivation from tumor dormancy. In the regression analysis, ER-positive breast cancers had a median time-to-death 3.67 years longer than ER-negative breast cancers (Table 2). However, this effect was attenuated by risk factors correlated with ER status (adjusted difference 1.76 years). Notable independent predictors of time-to-death include ethnicity, tumor size, grade, nodal status and PR-status. Conclusion: Reactivation from tumor dormancy appears to occur at random wherein the rate of tumor activation is correlated with probability of death from breast cancer. This model may explain the very long time to cancer recurrence in women with ER-positive node-negative breast cancers. The clinical course of women with low-risk breast cancer is inherently unpredictable. Table 1: Factors related to breast cancer mortality and time-to-death by decile.Risk DecileNumber of patients (N)Annual death rate (per 100 person-years)20-year actuarial mortality (%)Peak mortality time (years)Median time-to-death (years)10th percentile time-to-death (years)90th percentile time-to-death (years)Optimal α (reactivations per person-year)112,3700.255.3%19.012.24.018.70.10212,3710.438.6%19.511.04.318.40.16312,3700.6011.3%9.510.33.517.60.20412,3710.7413.2%5.58.93.016.70.32512,3700.9016.0%3.58.62.617.40.32612,3711.1720.0%4.08.12.616.90.35712,3711.5124.8%3.07.62.316.70.41812,3702.1231.7%3.06.41.915.90.74912,3713.2242.5%2.55.61.714.81.291012,3706.6463.0%2.03.81.112.3InF (Ref.) Table 2: Predictors of time-to-death in breast cancer patients.PredictorValueUnadjustedAdjustedMedian time-to-death (years)Difference* (years)Difference (years)**POverall/ Reference5.6713.05***<.0001Year of diagnosis19905.42Reference19915.670.250.170.360919925.670.250.220.289919935.580.170.350.059919946.000.580.530.0054Age at diagnosis<505.58Reference50-595.920.330.070.580760-696.170.58-0.060.655170-805.25-0.33-1.18<.0001EthnicityWhite5.75ReferenceBlack4.33-1.42-0.320.009East Asian6.670.920.930.0058Southeast Asian6.080.330.090.8208Other/Unknown6.170.420.420.4176Tumor size (cm)<1cm8.33Reference1-2cm7.33-1.00-0.650.07922-3cm5.58-2.75-1.42<.00013-5cm4.75-3.58-1.76<.00015+cm3.83-4.50-2.16<.0001Tumor gradeI11.17ReferenceII7.58-3.58-2.82<.0001III4.25-6.92-4.78<.0001IV3.92-7.25-5.18<.0001Unknown6.25-4.92-3.57<.0001Nodal statusN06.83ReferenceN15.67-1.17-0.69<.0001N25.00-1.83-1.08<.0001N33.83-3.00-2.03<.0001Unknown2.42-4.42-2.32<.0001ER statusPositive7.17ReferenceNegative3.50-3.67-1.76<.0001PR statusPositive7.25ReferenceNegative3.83-3.42-1.7<.0001Unknown5.08-2.17-1.35<.0001RadiotherapyNo5.42ReferenceYes6.080.670.260.0214Unknown5.08-0.33-0.470.1043ChemotherapyNo/Unknown6.58ReferenceYes5.17-1.42-0.310.0091*Difference in median time-to-death relative to reference value.**Independent difference in median time-to-death after adjusting for all covariates in table.***Median time-to-death of the reference group. Citation Format: Vasily Giannakeas, Steven A Narod. A generalizable relationship between mortality and time-to-death among breast cancer patients can be explained by tumor dormancy [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P5-07-07.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.010 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.008 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".