AS370 - WHI cancer survivor cohort (LILAC)

Investigator Names and Contact Information

Garnet Anderson (garnet@whi.org)

Electra Paskett (OSU), Bette Caan (Kaiser)

Introduction/Intent

Among women between the ages of 35 and 74, cancer takes a larger fraction of lives than any other condition [1] and is associated with tremendous morbidity and cost. With the growing population, particularly in older age‐groups, cancer‐related medical expenditures are projected to reach at least $158 billion in 2020 – an increase of 27 percent from 2010 after controlling for inflation and may reach $207 if trends in costs for newly developed diagnostic, treatment and monitoring technologies persist [2]. With increasing survival has come a keen awareness of the longer‐term impacts of the disabilities associated with cancer and its treatment. The need to identify ways to reduce the burden of cancer, through prevention, early detection, and improved, targetted therapies remains an urgent element of the nation’s health research agenda.

Increasingly cancer treatment and etiological studies are based on the molecular characterization of the tumor. Breast cancer provides a pertinent example. Understanding the role of the estrogen receptor led to an array of treatments (selective estrogen receptor modulators and aromatase inhibitors) that interfere with the estrogen pathway and improve prognosis for women with ER+ disease. Similarly the discovery of HER2/nue and its role in the EGF pathway led to the development of trastuzumab, an effective therapy for the fraction of women with HER2 positive disease. Outside of breast cancer, the discovery of the Philadelphia chromosome mutation and the hyperactive bcr‐abl protein soon led to the development of imatanib, a tyrosine kinase inhibitor shown to be effect first in CML and subsequently in GIST. By defining molecular subgroups of disease, treatment is gradually becoming more personalized— new therapies are being identified that have an established role in the biologic pathway of disease and patients without the designated tumor phenotype are spared the cost and side effects of treatments unlikely to provide benefit.

Molecularly defined subgroups may also be important in improving our understanding of disease etiology and risk. Heterogeneity in breast cancers is now more universally acknowledged. Classification into luminal A, luminal B, triple negative/basal‐like, or HER2 positive subtypes, based on the ER, PR, HER2, Ki67, cytokeratin 5/6 and/or HER1 expression is quickly becoming the gold standard. Future studies of breast cancer risk factors need to take into account these subtypes. For example, traditional breast cancer risk factors associated with reproductive history (e.g., age at menarche, first birth, breast feeding) and weight , have much stronger association with risk of both luminal A and luminal B (hormone receptor positive) tumors suggesting that triple‐negative or basal phenotype tumors may have a distinct etiology[3]. The molecular characterization of most other cancer sites is even earlier in the development process. The example from breast cancer suggests that while initial discovery work may require fresh tumor specimens, technology eventually develops rapidly to allow adequate tumor phenotyping on fixed tissue[4], making the development of a repository of such tissue integrated with a database of epidemiological and clinical data an extremely valuable resource. Further, pairing such specimens with blood specimens collected months or years prior to diagnosis would create a straightforward pipeline of translational research from molecular characterization of tumors to work in clinically relevant specimens for risk and early detection.

The Women’s Health Initiative (WHI) has made some important contributions to cancer research and is well‐poised to do more. The practice‐changing results of the WHI estrogen plus progestin trial [5‐7] led to a rapid reduction in hormone therapy use [8] that was followed by a reduction in breast cancer incidence observed both nationally [6,9] and internationally[9‐14]. Although less recognized, results from the other WHI randomized trials have revealed important information on cancer. Use of estrogen alone was shown to reduce the risk of breast cancer [15‐17], especially when initiated later in life [18]. In the low‐fat diet trial of 48,000 women, WHI observed a 9% (non‐significant) reduction in breast cancer incidence was observed over a mean follow‐up of 8 years, with a suggestion of larger benefit for PR‐ tumors [19], and a 17% reduction in ovarian cancer risk [20] but no effect on colorectal cancer rates [21]. Similarly, the randomized trial of calcium and vitamin D supplements found no effect of these supplements on risk of colorectal cancer [22], breast cancer [23] or total cancer incidence [22]. In addition to these trials, the extensive information on lifestyle and environmental exposures as well as pre‐diagnositic blood collection within the larger WHI cohort has supported many epidemiologic studies of risk factors for a variety of cancers, more mechanistic studies of genomic and proteomic factors, and early detection.

No information has been systematically collected by WHI on cancer treatment or its outcome, limiting the ability of WHI to address issues of prognosis and long‐term effects in this otherwise wellcharacterized cohort. In addition, changing concepts in tumor phenotyping are making somewhat obsolete the current WHI standard of coding cancers simply by SEER categories of histology, stage, grade and tumor size. For both etiological and therapeutic purposes, cancers are being characterized by molecular subtypes assessed through gene or protein expression. Thus while WHI has many strengths in its existing infrastructure for supporting cancer studies in both the numbers of cancer cases captured and the annotation with high quality epidemiologic factors and pre‐diagnostic specimens, the inability to classify these tumors by modern methods will increasingly limit the information that can be derived from these studies.

Specific Aims

In this cancer cohort infrastructure proposal, we seek to develop the WHI program into a more comprehensive resource for cancer research for molecular epidemiological and cancer survivorship studies by addressing the two key limitations in the existing program. Specifically, we propose:

• 2.1 To collect information on cancer treatment and outcomes, including recurrence, progression and long‐terms effects of treatment in women diagnosed with cancer (other than non‐melanoma skin cancer) during their WHI participation (past or future), using CMS (Medicare) linkage where applicable, and direct medical record abstraction otherwise.

• 2.2 To expand the WHI biorepository of fasting serum, plasma, and DNA collected prior to diagnosis with paraffin embedded tumor tissue for selected cancers sites, and through this process to develop the logistical and administrative mechanisms to support additional, targeted tissue collection as needed and funding permits.

• 2.3 To conduct methodologic research aimed at improving the efficiency and validity of data collection and analysis, including

  • 2.3.1 Exploring the opportunities, mechanisms and costs of collecting cancer treatment and outcome data through linkage to other administrative databases such as SEER and the HMO research network (HMORN).

  • 2.3.2 Developing and refining algorithms and standards for extracting cancer treatment and outcomes data from administrative databases

  • 2.3.3 Assessing the comparability/validity of cancer treatment and outcome data collected by linkages to administrative databases compared to those obtained from direct medical record review for a subset of primary sites.

This enhancement to WHI will facilitate research on the etiology and risk factors for cancer incidence, cancer recurrences and survival, with a particular emphasis on the contribution of co‐morbidities, lifestyle factors, molecular and genetic factors, and treatment‐related factors (ie type of treatment and adherence) that decrease the risk of second and recurrent cancers. In addition, we will be able to examine the effects of cancer diagnoses and treatment on overall health and quality of life within the larger WHI cohort. By creating a repository that will allow up‐to‐date phenotyping methods to occur on a large number of tumors from many cancer sites, the WHI will be in a better position to contribute to the advancement of our understanding of the molecular basis of the disease and its treatment for the foreseeable future.