Figure 1 displays the study attrition. 221,221 records were examined that had evidence of all three receptors’ (ER, PR and HER2) results, of which 24,722 (11.1%) were found to have TNBC, and 8,409 patients (34.0%) were examined during the study period of September 1st, 2016 to September 30, 2021. Of these 8,409 patients, 7,470 (88.8%) had two or more office visits, were accessible for research, and were over age 18. Finally, 3,321 (44.4%) patients had stage IIa, IIb, IIIa, or IIIb cancer during the study period and had complete data in the iKM EHR database. These 3,321 patients formed the cohort used for examination of neoadjuvant therapy administrations. A subset of this population of 3,321 patients who were treated at practices with GREAT programs formed the GT cohort of 1,318 patients (39.7%).
Figure 1 Patient selection by serial attrition according to protocol inclusion and exclusion criteria
Figure 2 shows the rates of GT and NC treatment according to various social determinants of health variables. By univariate analysis, Figure 2A shows that patients who received GT were significantly (p < 0.05) younger, less likely to be Black, more likely from larger practice sizes, less likely to have Medicare or self-pay, and resided in a higher socioeconomic status region (as indicated by lower ADI). Specifically, those tested had a mean age of 48.4 years, while those not tested had a mean age of 50.7 years. The percent tested was 77%, 68%, 75%, and 77% for Asian, Black, White and other. The percent tested in each US geographic region was 70%, 83%, 68%, and 87% for midwest, northeast, south and west, respectively. The mean practice size was 32.8 physicians / practice for those tested, and 29.2 physicians / practice for those not tested. Notably, the ADI was 42.3 for those tested and 47.1 for those not tested.
Figure 2B shows that those using NC were significantly (p < 0.05) younger (mean of 55.9 years, vs. 61.8 years), less likely to have a normal BMI (in either direction; percent receiving NC for normal: 65%; underweight: 73%; overweight 69%; obese: 71%), more likely to be treated in the western region (percent receiving NC for west: 69%, midwest: 65%, northeast: 63%, and south: 61%), more likely to come from larger practice sites (those receiving NC had mean 28.5 physicians/practice, those not receiving NC 24.9 physicians/practice) and not use Medicare (percent receiving NC among those having commercial insurance: 70%, Medicaid: 71%, Medicare: 54%, and self-pay: 64%). Notably, the ADI and race distribution was similar between those using and not using NC (mean ADI was 40.5 among those receiving NC and 40.7 among those not receiving NC; for all race groups, percent receiving NC ranged from 60%-66%).
Figure 2 a. Germline genetic testing (n=1318 totally) or b. neoadjuvant chemotherapy administration (n=3321 totally), according to social determinants of health category were significantly different (*, p < 0.05) different for the group. Percentages of the group receiving NC or genetic testing are shown at the end of the bar, and sample sizes receiving NC or genetics tests, as a fraction of total for the subgroup are shown within bars
Figure 3 compares those that did or did not receive GT, segregated by covariates of health and social characteristics, via univariate ANOVA (Figure 3A, left), multivariate analysis (Figure 3A, right), and a multivariable comparison of odds ratios (Figure 3B). Via multivariate ANOVA or odds ratio comparison, those less likely to receive GT were Black (p = 0.014; compared to White via ANOVA and p = 0.007; OR = 0.43; compared to White via comparison of odds ratios) and those with a higher ADI (p = 0.001 via multivariable ANOVA and p = 0.001; OR = 0.98 via comparison of odds ratios). No other social or clinical characteristic analyzed revealed differences between patients who had GT versus those who did not.
Figure 3 Germline genetic testing analyzed by contribution of social determinant of health. For each characteristic, the first one listed is the control group and bold font indicates significant change
In Figure 4, using the same set of analyses, we then looked at patients who did or did not receive NC segregated by the same covariates of health and social characteristics as used in Fig 3. Via multivariable ANOVA analysis or odds ratio comparison, those less likely to receive NC were those on Medicaid (compared to commercial insured; p = 0.026 via multivariable ANOVA in Figure 4A, right; p = 0.02 and OR = 0.54 via odds ratio in Figure 4B), older patients (p < 0.001 via multivariable ANOVA in Figure 4A, right; p < 0.001 and OR = 0.95 via odds ratio in Figure 4B). Those more likely to received NC were Black (compared to Whites; p = 0.033 via multivariable ANOVA in Figure 4A, right; p = 0.044 and OR = 1.57 via odds ratio in Figure 4B), stage IIB (p < 0.001 via multivariable ANOVA in Figure 4A, right; p < 0.001 and OR = 2.32 via odds ratio in Figure 4B), stage IIIA (p < 0.001 via multivariable ANOVA in Figure 4A, right; p < 0.001 and OR = 2.59 via odds ratio in Figure 4B) and stage IIIB/C (p < 0.001 via multivariable ANOVA in Figure 4A, right; p < 0.001 and OR = 4.08 via odds ratio in Figure 4B) (all stages compared to stage IIA), patients treated in the western region of the US (compared to northeast patients; p = 0.024 via multivariable ANOVA in Figure 4A, right; p = 0.026 and OR = 2.49 via odds ratio in Figure 4B) and obese patients (compared to normal BMI patients; p = 0.037 via multivariable ANOVA in Figure 4A, right; p = 0.048 and OR = 1.46 via odds ratio in Figure 4B).
Figure 4 Neoadjuvant chemotherapy, analyzed by contribution of social determinant of health. For each characteristic, the first one listed is the control group and bold font indicates significant change
Table 1 summarizes significant factors from Figure 3 and 4 that affect use of NC or GT. Importantly, no one factor, except for treatment in the western part of US, affected NC and GT in the same direction.
Table 1. Summary of socioeconomic factors significantly associated with the use of germline testing and neoadjuvant chemotherapy for TNBC, via odds ratio comparison (refer to Figure 3B and 4B)
Multivariable analysis results
|
Statistically significant (p < 0.05)
Clinical Outcome
|
|
Germline Testing
|
Neoadjuvant Chemotherapy
|
Factors associated with higher utilization
|
Treated in the west (p = 0.049)
|
Treated in the west (p = 0.026)
Black race (p = 0.044)
Obese category of BMI (p = 0.048)
Stage > IIA (p < 0.001)
|
Factors associated with lower utilization
|
Black race (p = 0.007)
Higher ADI score (p = 0.001)
|
Increased age (p < 0.001)
Insurance of Medicaid (p = 0.020)
|