br Statistical analysis br We first verified that
2.3.4. Statistical analysis
We first verified that the intervention was associated with eu-daimonic and hedonic well-being, depressive symptoms, and CTRA gene expression using paired samples t-tests to test the significance of pre-post intervention change. As in previous studies (e.g., Antoni et al., 2016), we assessed change in expression of the overall 53-gene CTRA SYBR Safe DNA Gel Stain score, which includes a proinflammatory component as well as its inverse component involving 31 gene transcripts involved in Type I interferon response (e.g., IFNB1, MX1, OAS1, IFIT1) and 3 gene tran-scripts involved in antibody synthesis (e.g., IGJ). Pro-inflammatory genes were weighted + 1 and antiviral and antibody-related genes were weighted −1 as in previous research (Cole et al., 2015; Fredrickson et al., 2013, 2015; Kitayama et al., 2016). Standard errors for the change scores were derived from bootstrap resampling of linear model residual vectors (controlling for correlation among genes). Ancillary analyses also verified whether this sample showed changes in bioin-formatically inferred activity of the pro-inflammatory transcription factor, NF-κB (using the TELiS system as previously described; Bower et al., 2015; Cole et al., 2005), and for preferential derivation of down-regulated gene transcripts from the pro-inflammatory CD16- “classical” subset of monocytes (using Transcript Origin Analysis as previously described; Bower et al., 2015; Cole et al., 2011). These ancillary tran-scriptome-wide analyses took as input all gene transcripts showing > 1.15-fold change in expression from pre- to post-intervention; this threshold is consistent with our prior work (e.g., Bower et al., 2014; see supplement page 1–2 for additional detail on genomic analyses).
Our primary analyses used linear regression to quantify associations between the magnitude of change in eudaimonic well-being and the magnitude of change in the global CTRA score. Significant bivariate associations were followed up with covariate-adjusted analyses con-trolling for age, BMI, time since cancer diagnosis, tumor stage, and use of endocrine therapy. To further interrogate the specificity of this as-sociation, we also tested whether this association held when simulta-neously controlling for change in hedonic well-being. Secondary ana-lyses were then conducted to examine whether increases in hedonic well-being alone were associated with the magnitude of change in the global CTRA score, and whether decreases in depressive symptoms alone were associated with the magnitude of change in the global CTRA score.
3.1. Sample characteristics
Women endorsed feelings of eudaimonic well-being two-to-three times a week over the past month (M = 3.11, SD = 1.01), with levels of
Demographic, Medical and Treatment-Related Characteristics of the Sample.
Employed full or part-time
Homemaker/volunteer 7 (32%)
Current endocrine therapy 9 (41%)
Note: M = mean; SD = standard deviation.
hedonic well-being in a similar range (M = 3.51, SD = 1.06). These levels of well-being are comparable to those observed in non-clinical samples (Lamers et al., 2010). Depressive symptoms were elevated at baseline; mean scores on the CES-D (M = 15.50; SD = 10.68; range 1–40) were comparable to other samples of younger breast cancer survivors (Ganz et al., 2012), including those in our previous inter-vention study (Bower et al., 2015). Of note, 55% endorsed clinically significant depressive symptoms, as indicated by CES-D scores greater than or equal to a 16.
3.2. Preliminary analyses: Intervention-associated changes in well-being, depressive symptoms, and gene expression
Paired samples t-tests demonstrated significant changes in well-being and depressive symptoms from pre to post-intervention (see Table 2). More specifically, there was a significant increase in eu-daimonic well-being (t(20) = 3.23, p = 0.004, d = 0.71) and hedonic well-being (t(20) = 2.16, p = .043, d = 0.47). Depressive symptoms decreased significantly after the intervention, (t(21) = −2.97, p = .007, d = 0.63), with an effect size similar to that observed in our previous RCT (d = .54; Bower et al., 2015). See Table 3 for correlations among psychosocial measures. In genome-wide transcriptional profiling of peripheral blood monocytes, there was no significant change from pre- to post-inter-vention in average expression of a 53-gene CTRA composite score
Table 2 r> Pre- and Post-Intervention Means and Standard Deviations for Eudaimonic Well-Being, Hedonic Well-Being, and Depressive Symptoms.
Outcome Mean SD Mean SD p value Cohen’s d
Correlations for Eudaimonic Well-Being, Hedonic Well-Being, and Depressive Symptoms.
Note: T1 = Pre-intervention; T2 = Post-intervention.