br Allele and genotype frequencies were determined
Allele and LY294002 frequencies were determined by gene counting. Hardy-Weinberg equilibrium was assessed by the x2 test. Measures of linkage disequilib-rium (ie, D’ and r2) were computed with Haploview 4.2. Linkage disequilibrium-based haplotype block defini-tion was based on the D’ confidence interval method.18
For SNPs that were members of the same haploblock, haplotype analyses were conducted to localize the associ-ation signal within each gene and to determine whether haplotypes improved the strength of the association with the phenotype. Haplotypes were constructed using the program PHASE version 2.1.60 To improve the stability of haplotype inference, the haplotype construction proce-dure was repeated 5 times using different seed numbers with each cycle. Only haplotypes that were inferred with probability estimates of ≥.85, across the 5 iterations, were retained for downstream analyses.69 Ancestry informative markers (AIMs) were used to
minimize confounding owing to population stratifica-tion.20,21,62 One hundred six AIMs were included in the
ARTICLE IN PRESS
4 The Journal of Pain
analysis. Homogeneity in ancestry among patients was verified by principal component analysis51 using Helix Tree (Golden Helix, Bozeman, Montana). The first 3 prin-cipal components (PCs) were used as covariates in the regression analyses to adjust for potential confounding owing to population substructure (ie, race/ethnicity).
Three genetic models were assessed for each SNP (ie, additive, dominant, recessive). The genetic model that best fit the data was selected for each SNP. Logistic regression analysis that controlled for significant covari-ates, as well as genomic estimates of and self-reported race/ethnicity, was used to evaluate the associations between genotype and pain group membership. A back-ward stepwise approach was used to create a parsimoni-ous model. Except for genomic estimates of self-reported race/ethnicity, only predictors with a P value of <.05 were retained in the final model. Genetic model fit and both unadjusted and covariate-adjusted odds ratios were estimated using StataSE version 14. Owing to the explor-atory nature of this study, a P value of <.05 was consid-ered significant.
Differences in Demographic and Clinical Characteristics Between Arm Pain Classes
Table 1 summarizes the significant differences in demographic and clinical characteristics between the no arm pain and mild arm pain classes. Compared with the no arm pain class, women in the mild arm pain class
Genetic Associations With Persistent Arm Pain
were significantly younger, had more education, had a lower KPS score, and were less likely to have high blood pressure. These women had a more advanced stage of disease, had a higher number of breast biopsies, had an axillary lymph node dissection (ALND), and had a greater number of lymph nodes removed. A greater per-centage of women in the mild arm pain class had pain in the breast before surgery, reported strange sensations in the affected breast, and had higher average and worst postoperative pain scores. These women were more likely to have had a surgical drain, had a greater number of drains, were more likely to have received neoadjuvant chemotherapy, and a higher percentage had received a biologic therapy during the 6 months after surgery.
Table 2 summarizes the significant differences in demographic and clinical characteristics between the no arm pain and the moderate arm pain classes. Compared with the no arm pain class, women in the moderate arm pain class were younger, had lower KPS scores and annual household incomes, a higher body mass index and Self-Administered Comorbidity Questionnaire scores, and were less likely to be white. A greater percentage of women in this class reported comorbid anemia; were less likely to have breast fed; had more advanced disease; reported breast pain before surgery; reported sensations of swelling, numbness, and hardness in the affected breast; had received neoadjuvant chemotherapy; had a greater number of breast biopsies; underwent a mastec-tomy; had a greater number of lymph nodes removed; had a surgical drain; had a greater number of drains
Table 1. Significant Differences in Demographic, Clinical, and Surgical Characteristics Between Women in the No Arm Pain (n = 164) and Mild (n = 93) Arm Pain Classes
CHARACTERISTICS NO PAIN MILD PAIN STATISTICS
Abbreviations: FE, Fisher’s exact; U, Mann-Whitney U test.
NOTE. Values are mean § standard deviation or percentage (number).
ARTICLE IN PRESS
Knisely et al The Journal of Pain 5