Exploratory Data Analysis on Breast Cancer Prognosis

Introduction

Breast cancer is the most common female cancer in the US, the second most common cause of cancer death in women (“American Cancer Society”, 2016), and the main cause of death in women ages 40 to 59 (Siegel et al., 2012).

In 2016, it is estimated that 249,260 new cases of breast cancer will be diagnosed and estimated 40,890 breast cancer deaths; and an invasive breast cancer will be diagnosed in about 246,660 women and 2,600 men. An additional 61,000 new cases of in situ breast cancer will be diagnosed in women (“American Cancer Society”, 2016). The lifetime probability of developing breast cancer is one in six overall (one in eight for invasive disease) (“American Cancer Society”, 2013; Siegel et al., 2012).

Worldwide, breast cancer is the most frequently diagnosed cancer among women in 140 of 184 countries, according to the World Cancer Research Fund International (“BCRF”, n.d.). Since 2008, breast cancer incidence has increased by more than 20 percent and mortality has increased by 14 percent (“BCRF”,n.d.). Nearly 1.7 million new breast cancer cases were diagnosed in the last report of 2012.

Medical prognosis is an evaluative component of medicine that encompasses the science of estimating the complication and recurrence of disease and predictive survival of patients (Ohno-Machado, 2001). Medical prognosis plays an increasing role in health care outcome. Reliable prognostic models that are based on survival analysis statistics and techniques have been applied to a variety of domains with varying degrees of success (i.e., APACHE IV (Zimmerman, 2006)).

Breast cancer prognosis is a vital element of providing effective treatment for patients. It has become increasingly important that clinicians are provided with accurate prognostic information on which to base therapeutic decision as the range of options for the treatment of patients with breast cancer widens. A large number of factors, including tumor grade, tumor size, and lymph node status including other aspects may influence or correlate with prognosis for breast cancer patients.

Breast cancer prediction survivability has mainly been studied through clinical approaches, based on pathological factors such as tumor grade, tumor size and number of the positive lymph nodes, estrogen (ER), progesterone (PR), and human epidermal growth factor receptor 2 (Her2) receptors, etc. Most studies are carried out in an effort to find factors that clarify the large unexplained variation in prognosis of the breast cancer patients. There is still uncertainty about the importance of most prognostic factors. There are other non-clinical factors such as age, ethnicity, obesity, and marital status that may have prognostic impact but are not used routinely in clinical practice.

This chapter is a survey of the significance of the non-clinical prognostic factors (i.e., age, ethnicity) in finding the prognosis for breast cancer patients. The National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Public-Use Data (years 1977-2013; 421,056 records) (“SEER,” n.d.) and the Howard University Cancer Center Tumor Registery (1995-2013; 1599 records) data were analysed. NPI (Nottingham Prognostic Index (Galea, 1992)) is a prognostic tool that enables grouping of patients based on calculated prognosis (i.e., excellent, good, moderate, poor) and the impact of each non-clinical prognostic factor on these subgroups. In addition, survival analysis tools such as Cox proportional hazards and Kaplan-Meier survival curve (“Cox proportional-hazards regression”, 2013) were used in analyzing the data.

Our data analysis suggested that age, ethnicity, and marital status have some influence on the survivability rate of breast cancer patients. Whether such influence is significant enough in relationship with the clinical factors such as tumor size and grade remains to be further studied.