Supplemental Table 1: Multivariate analysis of variables potentially affecting overall (i.e., all-cause) survival in women with breast cancer (BC) who subsequently developed histologically confirmed non-small cell lung cancer (NSCLC)

Localized
NSCLC ** / Regional
NSCLC ** / Distant
NSCLC **
No. BC-NSCLC / 1215 / 1042 / 1272
No. NSCLC / 35,075 / 41,512 / 77,528
Breast cancer variables
Stage (regional)
p value / 0.680 / 0.492 / 0.599
 [SE] / -0.044 [0.103] / 0.062 [0.090] / 0.039 [0.077]
Estrogen receptor (neg/unknown)
p value / 0.221 / 0.065 / 0.0005
 [SE] / -0.114 [0.090] / -0.154 [0.082] / -0.224 [0.065]
Mastectomy (no/unknown)
p value / 0.166 / 0.105 / 0.813
 [SE] / 0.163 [0.120] / 0.155 [0.105] / -0.019 [0.082]
Radiotherapy (no/unknown)
p value / 0.998 / 0.372 / 0.979
 [SE] / 0.0003 [0.119] / -0.086 [0.103] / 0.002 [0.080]
NSCLC variables
Age at diagnosis
p value / < 0.0001 / < 0.0001 / < 0.0001
 (per decade) [SE] / 0.290 [0.049] / 0.253 [0.043] / 0.237 [0.032]
Laterality (contralateral/unknown)
p value / 0.057 / 0.933 / 0.177
 [SE] / 0.166 [0.086] / 0.007 [0.078] / 0.085 [0.064]
Histology (not adenocarcinoma)
p value / <0.0001 / 0.086 / 0.067
 [SE] / -0.401 [0.092] / -0.147 [0.082] / -0.121 [0.066]
Grade (lower)
p value / 0.517 / 0.004 / 0.0009
 [SE] / -0.069 [0.101] / -0.282 [0.095] / -0.281 [0.095]
Supplemental Table 1: (continued)
Localized
NSCLC / Regional
NSCLC / Distant
NSCLC
Latency
p value / 0.870 / 0.007 / 0.021
 (per year) [SE] / -0.002 [0.012] / -0.028 [0.010] / 0.016 [0.007]
Surgery (no/unknown)
p value / <0.0001 / <0.0001 / <0.0001
 [SE] / -1.297 [0.116] / -1.007 [0.096] / -0.915 [0.140]
Radiotherapy (no/unknown)
p value / 0.011 / 0.298 / 0.662
 [SE] / -0.353 [0.131] / -0.097 [0.088] / -0.030 [0.066]
Demographics variables
Race (non-white)
p value / 0.028 / 0.109 / 0.120
 [SE] / 0.347 [0.156] / 0.204 [0.121] / 0.145 [0.091]
County level education & income score *
p value / 0.009 / 0.186 / 0.077

** Each model is the result of stratified Cox proportional hazards regression analysis. Models included all indicated variables, plus adjustments for year of lung cancer diagnosis (stratified into a binary variable of: 1998 or earlier versus 1999 or later) and for county of residence. For county of residence, models adjust for the percentage of residents with less than a high school education and cost of living adjusted median household income as described below.* Baseline values of covariates are denoted in parentheses next to the variable name. Age and latency are coded as continuous variables, centered at their respective sample mean values, with larger values denoting greater time values.Regression parameter estimates capture the change per decade in age or year of latency, respectively

* As the education and income variables were found to be highly negatively correlated (r=-0.71 in each stage group), we replaced these predictors with a univariate socioeconomic status score that was derived from the education and income variables using principal component analysis(accounting for 86% of total variance). This score was found to be highly positively associated with the proportion of adults residing within that county age ≥25 years with less than a high-school education (r=0.92) and highly negatively associated with the cost-of-living adjusted median income (r=-0.92).We report only the p-value for this derived variable.

Abbreviation: SE=standard error

Supplemental Table 2: Multivariate analyses of variables potentially affecting cause-specific survival and mortality incidence in women with breast cancer (BC) who subsequently developed histologically confirmed non-small cell lung cancer (NSCLC)

Localized NSCLC ** / Regional NSCLC ** / Distant NSCLC **
No. BC-NSCLC / 1,215 / 1,042 / 1,262
No. NSCLC / 33,860 / 41,512 / 76,256
Cumulative incidence / Lung cancer CSS / Cumulative incidence / Lung cancer CSS / Cumulative incidence / Lung cancer CSS
Breast cancer variables
Stage (regional)
p value / 0.276 / 0.257 / 0.023 / 0.017 / 0.003 / 0.009
 [SE] / 0.173 [0.159] / 0.176 [0.155] / 0.269 [0.118] / 0.281 [0.118] / 0.282 [0.094] / 0.248 [0.095]
Estrogen receptor (neg/unknown)
p value / 0.314 / 0.581 / 0.587 / 0.274 / 0.467 / 0.013
[SE] / 0.139 [0.138] / 0.073 [0.133] / -0.057 [0.105] / -0.112 [0.102] / -0.054 [0.074] / -0.190 [0.108]
Mastectomy (no/unknown)
p value / 0.405 / 0.309 / 0.333 / 0.188 / 0.002 / 0.021
[SE] / 0.145 [0.175] / 0.178 [0.175] / 0.122 [0.125] / 0.171 [0.130] / -0.288 [0.093] / -0.224 [0.097]
Radiotherapy (no/unknown)
p value / 0.585 / 0.526 / 0.392 / 0.326 / 0.013 / 0.010
 [SE] / -0.093 [0.171] / -0.111 [0.175] / -0.106 [0.124] / -0.125 [0.128] / -0.222 [0.089] / -0.155 [0.094]
NSCLC variables
Age at diagnosis
p value / 0.002 / < 0.0001 / < 0.0001 / < 0.0001 / < 0.0001 / < 0.0001
 (per decade) [SE] / 0.226 [0.072] / 0.303 [0.074] / 0.252 [0.054] / 0.319 [0.054] / 0.172 [0.037] / 0.288 [0.039]
Laterality (contralateral/unknown)
p value / 0.115 / 0.246 / 0.494 / 0.568 / 0.659 / 0.726
 [SE] / 0.203 [0.129] / 0.136 [0.117] / 0.066 [0.096] / 0.051 [0.090] / 0.028 [0.062] / 0.017 [0.050]
Histology (not adenocarcinoma)
p value / 0.002 / 0.0001 / 0.017 / 0.013 / 0.007 / 0.003
 [SE] / -0.434 [0.137] / -0.516 [0.134] / -0.248 [0.104] / -0.252 [0.101] / -0.198 [0.074] / -0.224 [0.076]
Grade (lower)
p value / 0.009 / 0.029 / 0.123 / 0.021 / 0.544 / 0.043
 [SE] / -0.410 [0.158] / -0.334 [0.153] / -0.183 [0.119] / -0.273 [0.118] / -0.057 [0.094] / -0.221 [0.109]
Latency
p value / 0.059 / 0.135 / 0.734 / 0.183 / 0.002 / 0.0004
 (per year) [SE] / 0.031 [0.017] / 0.024 [0.016] / -0.004 [0.013] / -0.017 [0.013] / 0.026 [0.008] / 0.030 [0.008]
Supplemental Table 2: (continued)
Localized NSCLC / Regional NSCLC / Distant NSCLC
Cumulative incidence / Lung cancer CSS / Cumulative incidence / Lung cancer CSS / Cumulative incidence / Lung cancer CSS
Surgery (no/unknown)
p value / 0.003 / <0.0001 / <0.0001 / <0.0001 / < 0.0001 / <0.0001
 [SE] / -0.517 [0.175] / -1.014 [0.168] / -0.668 [0.116] / -0.993 [0.119] / -0.564 [0.152] / -0.890 [0.170]
Radiotherapy (no/unknown)
p value / 0.992 / 0.223 / 0.096 / 0.623 / 0.263 / 0.716
 [SE] / - 0.002 [0.194] / -0.226 [0.186] / 0.185 [0.111] / 0.053 [0.107] / 0.083 [0.074] / 0.028 [0.078]
Demographic variables
Race (non-white)
p value / 0.184 / 0.102 / 0.010 / 0.013 / 0.021 / 0.012
 [SE] / 0.299 [0.225] / 0.383 [0.235] / 0.412 [0.161] / 0.406 [0.163] / 0.247 [0.107] / 0.282 [0.113]
County level education & income score *
p value / 0.626 / 0.148 / 0.332 / 0.880 / 0.975 / 0.404

** Regression parameter (), standard error (SE) and p-value for lung-cancer cause-specific mortality rate analyses derived from Cox proportional hazards models that includes same variables and stratification as in Table 4. Regression parameter () and p-value for cumulative incidence analysis is derived from the proportional sub-distribution hazard model introduced in Fine and Gray (1999) using the same variables and stratification as in the cause-specific analysis for lung cancer mortality. Baseline values of covariates denoted in parentheses next to variable name (age and latency are coded as for Table 4; see Table 4 caption for details).

* As the education and income variables were found to be highly negatively correlated (r=-0.71 in each stage group), we replaced these predictors with a univariate socioeconomic status score that was derived from the education and income variables using principal component analysis(accounting for 86% of total variance). This score was found to be highly positively associated with the proportion of adults residing within that county age ≥25 years with less than a high-school education (r=0.92) and highly negatively associated with the cost-of-living adjusted median income (r=-0.92).We report only the p-value for this derived variable.

Abbreviations: SE=standard error