Only the Lonely: Expression of Proinflammatory Genes Through Family Cancer Caregiving Experiences

Abstract

Objective.

The experience of cancer not only elicits turmoil but also resilience in the family, which has been related to psychological adjustment and physical health of family caregivers. The biological pathways linking family cancer caregiving to health, however, remain poorly understood. This study examined the extent to which psychological risk and resilience factors related to a proinflammatory gene expression profile (Conserved Transcriptional Response to Adversity: CTRA) among caregivers during the first year post-diagnosis of a colorectal cancer patient.

Methods.

A total of 41 caregivers (mean age=54 years old, 74% female, 40% Hispanic) provided psychological data and peripheral blood samples around 4 months and 12 months post-diagnosis. Mixed regression models controlling for demographic and biometric factors were used to test the associations of caregiver CTRA gene expression with caregiving stress, loneliness, and lack of social support (risk factors), as well as benefit finding and meaning (resilience factors).

Results.

When individually tested, all but benefit finding were significantly related to CTRA (R²≥.112, p<.045). When adjusted for other factors in either the risk or resilience group, loneliness, social support, and meaning effects remained significant (R²≥.120, p<.041). When all study factors were simultaneously adjusted (R²=.139), only loneliness remained significant (p=.034).

Conclusions.

Findings suggest that caregiving-related transcriptional effects appear to be most pronounced when caregivers experience low social support and loneliness, as well as little meaning or purpose in their caregiving. These findings suggest that the development of new intervention strategies that prioritize reductions in caregiver loneliness may favorably impact biological mechanisms related to caregiver health.

INTRODUCTION

Over 17 million individuals with a history of cancer were alive in the United States in 2019 and that number is estimated to grow to 22 million by 2030 (1). The majority of cancer survivors (67%) were diagnosed 5 or more years ago, which positions cancer as the second most prevalent chronic disease (2). Long-term mental and physical sequelae of cancer treatment, however, have imposed various challenges not only to the patients but also to their family caregivers (3–5). Elevated psychological distress among family cancer caregivers, which often exceeds that of the patients (6,7), has now been well documented. Further, evidence supporting increased risk for developing physical morbid conditions among caregivers has been accumulating in recent years (8–10). On the other hand, the turmoil cancer brings to the family has also been associated with personal growth and meaning finding experiences among family caregivers, which contribute significantly to the caregivers’ psychological adjustment (11–13). However, the contribution of personal growth and meaning to caregivers’ physical health remains unknown.

One pathway through which family cancer caregiving might impact the caregivers’ health involves inflammatory responses. Growing evidence from human social genomics research (14) has documented that exposure to various adverse conditions for extended periods of time is associated with a distinct shift in the basal gene expression profile of immune cells known as the Conserved Transcriptional Response to Adversity (CTRA). The CTRA expression profile is characterized by the up-regulation of proinflammatory genes (e.g., IL1B, IL6, IL8, and TNF) and down-regulation of genes involved in Type I interferon antiviral response (e.g., IFI-, OAS-, and MX- family genes) and antibody synthesis (e.g., IGJ; 14–16). Elevated CTRA profiles have been seen in individuals exposed to range of adverse life circumstances such as bereavement (17), chronic loneliness (18,19), chronic stress, including caregiving stress (20–22), poverty (23,24), racial discrimination (25), and trauma (26,27). As cancer caregiving involves both direct demands on the caregiver’s time, attention, and well-being and an extended period of distress associated with potential bereavement, elevated CTRA or up-regulation of proinflammatory gene expression is likely to be observed among family cancer caregivers. As elevated CTRA profiles have been associated with disease outcomes in patient and healthy populations (14–16,28–32), the elevated CTRA profile may be a biological pathway to compromised health of family cancer caregivers.

In contrast, reduced (down-regulated) CTRA gene expression has been seen in people who report high levels of a sense of purpose, meaning, and self-realization in their life (collectively referred to as eudaimonic well-being) and those who report or are engaged in self-transcendent prosocial behaviors (33–38). Prosocial or helping behaviors have also been associated with decreased stress physiology (39–41) and increased activity of neurobiological caregiving circuity (39,42–44), as well as reduced risk of cardiovascular disease and mortality (45–47). Therefore, reduced CTRA or down-regulation of proinflammatory gene expression is likely to be observed among family members providing help to their relatives with cancer, and such behaviors may have protective effects against development of morbidities.

These two groups of findings are both pertinent to caregiving but have seemingly contradictory implications. Caregiving for family members is a core example of prosocial behavior, which is often other-focused and self-transcendent, as opposed to self-focused. However, family caregiving, particularly in the medical illness context, is also clearly stressful and demanding, and often associated with chronic threat of bereavement or social loss (48,49).

A couple of observations may help disentangle these discrepant implications and potentially clarify the best ways to support caregivers’ health. One is that many of the studies linking prosocial behaviors to reduced CTRA involved (at least in part) prosocial behaviors directed toward strangers and suggest purpose and meaning as key psychological pathways linking prosocial behaviors to healthy gene expression. Helping family members with a major disease should be as purposeful and meaningful as helping strangers, if not more. Thus, the biological salutary effects of prosocial, helping behaviors and the associated psychological phenomena of feeling purpose and meaning, may be most apparent when in the absence of substantial stress or threat. On the other hand, most studies linking caregiving to morbidity and mortality have focused on caregiving stress, rather than meaning and purpose of caregiving. Thus, caregiving-related stress biology may be most pronounced when caregivers experience little meaning or purpose in their caregiving and are most attuned to its burdens and costs to their own well-being.

This study explored the extent to which 1) caregivers’ upregulated CTRA gene expression was associated with psychological risk factors, such as caregiving stress, loneliness, and lack of social support; and 2) caregivers’ down-regulated CTRA gene expression was associated with psychological resilience factors, such as purpose and benefit-finding from caregiving. We followed a cohort of 41 family caregivers of patients diagnosed with colorectal cancer from around the time of diagnosis and treatment initiation to about one-year later.

Methods

Participants and Procedures

Adult cancer patients and their family caregivers were invited to participate in a psychobiological observational study (50). Cancer patients who were newly diagnosed with colon or rectal cancer (stage I-IV, > 21 years old) less than four months prior to participating in the study (T1) were recruited at the University of Miami and Northwestern University oncology clinics. Patients identified family members or individuals considered family who were providing unpaid help during their cancer experience (e.g., providing emotional support or medical information, paying for groceries, transportation to clinic). Eligibility criteria for family caregivers were 21 years of age or older, self-identified as Black, Non-Hispanic White, or Hispanic, and speaking and reading proficiency in English or Spanish at the 5^th grade level. Exclusion criteria included active/untreated psychosis, substance abuse/dependence, and suicidal ideation within the past year; and for blood sample, HIV seropositive status. A follow-up assessment was conducted approximately one-year post diagnosis of the patient (T2: M = 11.6 months post diagnosis, SD = 1.34 months).

This study was approved by the University of Miami and Northwestern University Institutional Review Boards. Caregivers completed a questionnaire assessing risk psychological factors, resilience psychological factors, and demographic and biobehavioral factors, alone at home or in the clinic; and non-fasting blood samples were drawn by study phlebotomist at both T1 and T2. We used translation-back-translation method for study materials in Spanish, which was approved by the University of Miami Institutional Review Board. Participating caregivers were provided a $40 incentive at each assessment. A total of 41 (24 from Miami site and 17 from Northwestern site; 13 completed the questionnaire in Spanish) caregivers provided complete data including questionnaire data and blood samples at either T1 or T2 (37 provided blood data at both T1 and T2). Participants who provided signed informed consent form were sent with follow-up introductory letter and received calls to schedule follow-up assessments that could be done at their home or in the clinic. Primary reasons for drop-out at T2 were “no longer interested in the study”, “no time for study/research”, and “do not want to think about cancer”.

Measures

Risk Factors.

Three aspects of risk psychological factors studied include caregiving stress, loneliness, and lack of social support. Caregiving stress, the extent to which caregivers felt overwhelmed by care tasks and responsibilities, was measured using the 4-item stress overload subscale of the Pearlin Stress Scale (51), with responses ranging from 1 = not at all to 4 = completely. An example item is “I work hard as a caregiver but never seem to make any progress.” The mean of the responses defined caregiving stress, with higher scores reflecting a greater perceived level of caregiving stress. The subscale had good internal consistency in the present study (α = .844). Loneliness, defined as perceived social isolation, was measured using the 4-item abbreviated UCLA Loneliness Scale (52,53), with responses ranging from 1 = never to 4 = always. The 4 items had good internal consistency (α = .809). Finally, social support, the extent to which the family caregivers perceived that emotional, informational, and instrumental support were available to them, was measured using the 6-item Interpersonal Support Evaluation List (ISEL: 54). The composite score was a mean of the 6 items, rated on a 4-point scale (1 = definitely false, 4 = definitely true), which had acceptable internal consistency (α = .838). Lower social support scores served as an indicator of greater stress. These three risk factors were moderately correlated with each other in the current sample (|.126| < r |.323|, p < .001), which suggested that they were conceptually related but somewhat distinct constructs (Table S1, Supplemental Digital Content).

Resilience Factors.

Two psychological resilience factors assessed were benefit finding and sense of meaning. First, benefit finding, the degree to which caregivers experienced personal growth or positive changes following the cancer in the family was measured using the 17-item Benefit Finding Scale (55,56). Instructions were modified so that participating caregivers rated each item in response to providing care for their relatives’ through their cancer experience, using a 5-point Likert-style format (1 = not at all, 5 = extremely). The composite score of the mean of the 17 items had good internal consistency in the present study (α = .977). Another construct assessed was meaning, or the extent to which caregivers found meaning in their lives through their cancer caregiving experience, which was assessed using the 4-item Meaning subscale of the Functional Assessment of Chronic Illness Therapy-Spiritual Well-being Scale (FACIT-Sp: 57,58). The meaning composite score (mean of the 4 items, on a 5-point Likert scale; 1 = Not at all, 5 = Very much) had good internal consistency in the present study (α = .951). These two resilience factors were moderately correlated with each other (r = .386, p < .001) and with two risk factors (loneliness and social support: |.142| < r |.459|, p < .001).) but not with caregiving stress (r < −.014, p > .44) in the current sample (Supplemental Table 1).

Transcriptome Profiling.

Blood samples were collected by venipuncture and peripheral blood mononuclear cells (PBMC) were isolated by standard Ficoll gradient centrifugation within 24 hours of sample collection. PBMC were stored at −80C and transported to the UCLA Social Genomics Core Laboratory, where RNA was extracted (Qiagen RNeasy), tested for suitable mass (Nanodrop ND1000) and integrity (Agilent Bioanalyzer), converted to cDNA libraries (Lexogen QuantSeq 3’ FWD), and sequenced on an Illumina HiSeq 4000 instrument in the UCLA Neuroscience Genomics Core Laboratory, all following the manufacturers’ standard protocols for this workflow. All samples were assayed in a single batch, and procedures targeted 10 million 65-nt single stranded sequencing reads per sample (achieved average 10.9 million), each of which was mapped to the consensus human transcriptome sequence using the STAR aligner (average 95.5% successfully mapped), and quantified as gene transcript counts per million total mapped reads.

The primary outcome analyzed in this study was the expression of an a priori-specified multi-gene composite score previously developed to assess the CTRA transcriptome profile (34,35). This composite involves average expression levels for 19 pro-inflammatory gene transcripts (e.g., IL1B, IL6, TNF, CXCL8/IL8) and 34 gene transcripts involved in Type I interferon antiviral responses and anti- body synthesis (e.g., JCHAIN and IFI-, MX-, and OAS-family genes). Gene expression values were floored at 1 transcript per million mapped reads and log2-transformed to mitigate skew and heteroscedasticity. To facilitate statistical model estimation, data for 9 genes showing minimal variability in expression level (SD > .5 log2 units) were screened out and data for the remaining 44 genes were z-score standardized within gene, with signs inverted for the interferon and antibody-related genes to reflect their inverse contribution to the CTRA profile (34,35).

Demographic and Biometric Factors.

Demographic and biometric factors that have been known to be associated with biological markers of stress or specifically with CTRA were considered to be included in the statistical analyses as covariates (14–27). Those were age, gender/sex, and ethnicity (Hispanic vs non-Hispanic reflecting the local/regional populations where the data were collected) at T1, current status of smoking and heavy alcoholic beverage drinking (consuming ≥ 3 alcoholic beverage drinks per day) (yes/no) at T1 and T2, which were self-reported. Height and weight at T1 and T2 were also self-reported, from which body mass index (BMI) was calculated (kg/m2). Also treated as a covariate was patients’ cancer stage at T1 (obtained from medical records or when not available, self-reported) coded as advanced cancer (stage III or IV) vs non-advanced cancer.

Statistical Analysis

Means and standard deviations or percentages of study variables are reported in Table 1. Since the prevalence of smoking (9.5%) and heavy alcoholic beverage drinking (6.8%) was skewed, these two variables were not included in the primary analyses but included in supplemental analyses. Zero-order correlation coefficients among study variables are reported in Supplemental Table 1. Mixed effect multivariate linear regression modeling was employed for the repeated-measures study design to test the primary inquiry on the extent to which risk and resilience psychological factors were associated with the CTRA profile around the time of diagnosis and treatment initiation (T1) and about one-year later (T2) among family caregivers. These models included fixed repeated measures effects for time and CTRA indicator gene and a random subject-specific intercept to control for interdependence among residuals within subjects. First set of analysis included covariates (age, gender/sex, ethnicity, BMI, and patient’s cancer stage) and assessment timepoint as fixed effects (we had no a priori hypothesis about any systematic change over time, but also did not randomly sample follow-up times or seek to generalize beyond the 2 time points examined here). Second sets of analysis included covariates, assessment timepoint, and one of the psychological risk or resilience factors (Table 2). Third sets of analysis included covariates, assessment timepoint, and either the three psychological risk factors or the two psychological resilience factors included as a set, to test mutually adjusted effects of either risk or resilience factors (Table 3). Finally, covariates, assessment timepoint, and all psychological risk and resilience factors were included in the equation to test mutually adjusted effects (Table 3). Supplemental exploratory analyses also tested the interaction effects of psychological factors with assessment timepoint. Statistical significance was set at a 2-tailed p-value < .05.

Table 1.

Sample Characteristics

	Mean (SD) or %	scale range
Age	54.04
Gender (Female)	74.4%
Ethnicity
Hispanic	39.7%
Non-Hispanic White	38.1%
African American	19.0%
Education
< High School	42.2%
College	40.0%
> Post-graduate	17.8%
Income
<$19,999	30.2%
$20,000–$39,999	20.9%
$40,000 – $75,000	11.6%
> $75,000	25.6%
Prefer not to answer	11.6%
Relationship to Patient
Spouse/Partner	33.3%
Offspring	24.4%
Friend	17.8%
Sibling	11.1%
Other	13.4%
Patients’ Cancer Stage
Stage I or II (Early)	48.6%
Stage III or IV (Advanced)	51.4%
BMI	30.92 (8.44)
Psychological Risk Factors
Caregiving Stress	1.91 (0.65)	1 – 4
Loneliness	2.05 (0.59)	1 – 4
Social Support	3.28 (0.58)	1 – 4
Psychological Resilience Factors
Benefit Finding in Caregiving	3.76 (0.85)	1 – 5
Spirituality-Meaning	3.20 (0.87)	1 – 5
CTRA	0.00 (0.99)	actual range: −6.269 – 5.379

N = 41 persons

Table 2.

Independent Associations of Psychological Risk and Resilience Factors with CTRA

	B	SE	t	p	95% CI	R2
Covariates						.086
Age	.00	.002	−0.069	.95	−.004 – .004
Gender (Male)	−.151	.067	−2.245	.028	−.285 - −.017
Ethnicity (non-Hispanic)	.027	.055	0.501	.618	−.082 – .137
BMI	−.004	.003	−1.342	.184	−.011 – .002
Patient’s non-advanced cancer	.010	.056	0.185	.85	−.100 – .122
Individual Psychological Factors
Caregiving Stress	.096	.047	2.058	.044	0.003 – 0.190	.120
Loneliness	.202	.067	3.009	.004	0.068 – 0.336	.122
Social Support	−.123	.049	−2.496	.015	−0.221 – −0.025	.115
Benefit Finding	.021	.037	0.574	.57	−0.053 – 0.095	.095
Spirituality-Meaning	−.073	.031	−2.349	.022	−0.135 – −0.011	.112

N = 41 persons with 2,904–3,212 observations; For effects of individual psychological factors, assessment timepoint, demographic and biometric factors (age, gender, ethnicity, BMI), and patients’ cancer stage were controlled for; R² is marginal, pseudo R² for repeated measures.

Table 3.

Simultaneous Associations of Psychological Risk and Resilience Factors with CTRA

	B	SE	t	p	95% CI
Psychological Risk Factors	ΔR2 = .056
Caregiving Stress	.056	.046	1.235	.22	−0.035 – 0.147
Loneliness	.140	.064	2.194	.032	0.012 – 0.267
Social Support	−.102	.048	−2.101	.040	−0.199 – −0.005
Psychological Resilience Factors	ΔR2 = .034
Benefit Finding	.066	.038	1.732	.088	−0.010 – 0.142
Spirituality-Meaning	−.097	.033	−2.899	.005	−0.163 – −0.030
All Psychological Factors	ΔR2 = .053
Caregiving Stress	.046	.046	0.998	.32	−.047 – .139
Loneliness	.140	.065	2.177	.034	.011 – .269
Social Support	−.062	.059	−1.041	.30	−.181 – .057
Benefit Finding	.052	.036	1.441	.16	−.020 – .124
Spirituality-Meaning	−.045	.038	−1.200	.24	−.121 – .030

N = 41 persons with 2,904–3,212 observations; Psychological factors were added to the equation to test their unique effects, controlling for assessment timepoint, demographic and biometric factors (age, gender, ethnicity, BMI), and patients’ cancer stage.; R² is marginal, pseudo R² for repeated measure

Results

Sample Characteristics

Caregivers were primarily middle-aged, female, Hispanic, relatively well educated, middle income, spouses of the index patient, providing care to patients whose cancer was diagnosed at an advanced stage, and on average overweight (Table 1). Overall, caregivers reported mild to moderate levels of feeling overwhelmed by caregiving and feeling lonely, while reporting moderate to high levels of perceived social support, finding benefit and meaning in caregiving.

Effects of Risk and Resilience Psychological Factors on CTRA Gene Expression

Before testing the extent to which individual risk or resilience factors associated with CTRA profile, we first examined whether CTRA gene expression profile changed significantly between T1 and T2. Controlling for covariates including demographic characteristics (age, gender/sex, ethnicity), BMI, and patient’s cancer stage, no significant change over time was observed, b = −.006, SE = .052, t = −.121, p = .90, 95% CI = −.111 – .098, suggesting the mean value of CTRA profile at T1 (.00025, SD = .9695) was not significantly different from that at T2 (−.00027, SD = 1.019). Thus, we treated the assessment time as a covariate in subsequent analyses.

Next, we examined the extent to which individual risk or resilience factors were associated with the CTRA profile controlling for covariates including demographic characteristics (age, gender/sex, ethnicity), BMI, patient’s cancer stage, and assessment time. As shown in Table 2, greater caregiving stress and loneliness were associated with upregulated CTRA gene expression (p < .045), whereas greater social support and meaning were associated with downregulated CTRA gene expression (p < .023). We observed no significant effect of benefit finding (p = .57).

When the three psychological risk factors were tested simultaneously to mutually adjusting each other’s effects (Table 3), only loneliness and social support, but not caregiving stress, remained significantly associated with CTRA gene expression (p < .041). When the two psychological resilience factors were tested simultaneously (Table 3), meaning remained significantly associated with CTRA gene expression (p = .005) and benefit finding became non-significant (p = .088).

Finally, when the three psychological risk factors and the two psychological resilience factors were tested simultaneously along with covariates, only loneliness continued to show a significant association with CTRA gene expression (Table 3; Figure 1).

Figure 1.

Associations of Psychological Risk and Resilience Factors with CTRA Gene Expression

Note. Data represent strengths of associations (B ± SE) of indicated predictors with the 44-gene CTRA indicator contrast in (A) separate analyses of individual psychological risk and resilience factors. (B) mutually adjusted analyses within psychological risk and resilience groups, and (C) mutually adjusted analyses of all psychological risk and resilience factor, in which each factor was adjusted for covariance with the other, adjusting for covariates (assessment timepoint, age, sex, ethnicity, BMI, and patients’ cancer stage).

We repeated the aforementioned analyses including smoking status and heavy alcoholic beverage drinking as additional covariates. The individual psychological risk factors remained significant but meaning became insignificant (t = 1.184, p = .24). When the psychological factors were tested in two groups, only loneliness among the psychological risk factor group was marginally significant (t = 1.963, p = .054), whereas within the resilience factor group, benefit finding became significant (t = 2.071, p = .042) and meaning became marginally significant (t = 1.855, p = .068). Finally, when all psychological factors were tested simultaneously, only loneliness remained significantly associated with CTRA gene expression (p = .032).

Although this study was not designed or powered to detect any moderating effects of time since patient diagnosis, we conducted supplemental exploratory analyses testing for possible interaction effects of psychological factors and assessment time. Results found no significant interactions for any psychological factor (p > .18), except benefit finding. The effects of benefit finding alone (B = .117, SE = .065, t = 2.582, p = .012, 95% CI =.038 – .296), or when adjusted for meaning (B = .152, SE = .069, t =2.20, p = .031, 95% CI = .014 – .289), or when mutually adjusted with all other psychological factors (B = .150, SE = .063, t =2.38, p = .021, 95% CI = .024 – .277) differed systematically across the two time points (Supplemental Table 2). Greater benefit finding was associated with downregulated CTRA gene expression at T1 and with upregulated CTRA gene expression at T2. However, it is important to note that these post-hoc interaction results would not reach significance after control for multiple testing across all the psychological factors examined in these exploratory interaction analyses.

Discussion

This study investigated the extent to which risk and resilience psychological factors were associated with CTRA gene expression among family caregivers of colorectal cancer patients during the first year since the diagnosis. Caregivers reported mild to moderate levels of psychological risk factors, such as caregiving stress, loneliness, and lack of social support, whereas they reported moderate to high levels of resilience characteristics, such as benefit finding and meaning. In other words, this sample of caregivers were in general relatively non-distressed and resilient during the first year since their relative’s diagnosis.

Keeping these general characteristics of the sample in mind, the substantial associations of individual variations in risk and resilience factors with the CTRA gene expression profile are quite remarkable. All but benefit finding were significantly associated with CTRA when examined individually, which is consistent with the existing findings on the detrimental biological impact of caregiving stress (20–22), loneliness (18,19), lack of social support (59,60), and inability to derive meaning from adversity (34,35,61).

Within the domain of stress, results from mutually adjusted analyses revealed that caregiving stress per se showed no significant unique association with CTRA gene expression after control for differences in loneliness and low social support, whereas the other two factors both continued to show significant associations, suggesting that perceived social isolation and lack of social support play a significant role in the overall relationship between the caregiver role and CTRA gene expression. This may have significant implications for the design of interventions to support caregiver health, as many interventions focus on ways in which caregiving stress can be reduced by providing educational skills training to caregivers to manage their patients’ symptoms effectively and to boost their esteem as a caregiver (62), and by employing cognitive behavioral (63,64) or mindfulness (65–67) techniques to reduce stress. A few stress management interventions for cancer caregivers have attempted to build social support and reduce loneliness (68,69). Our findings suggest these latter approaches may be more efficacious to mitigate the biological burden of caregiving for family members with cancer.

On the other hand, among the resilience factors we studied, caregivers’ experience of finding meaning in their lives through their cancer caregiving experience (meaning) was significantly associated with reduced CTRA, whereas the experience of personal growth or positive changes following the cancer in the family (benefit finding) was the case only around the time of diagnosis and treatment initiation but such experience was associated with elevated CTRA about one year after the diagnosis (i.e., the benefit finding x time point interaction). The favorable association with meaning is consistent with previous research linking reduced CTRA gene expression to high levels of a sense of eudaimonic well-being (purpose, meaning, and self-realization in life) and engagement in self-transcendent prosocial behaviors among healthy, non-caregiving adults (e.g., 34,37,38). Our findings extend this body of research to a clinical sample of family cancer caregivers and underscore the evolving nature of adjustment to medical illness in the family.

These findings may also provide guidance in the design of health-protective interventions for family cancer caregivers in emphasizing the critical role of meaning and social or spiritual significance of caregiving, rather than emphasizing an effort to identify personal growths emerging from cancer. This distinction may be a nuanced yet critical component that will shield against the biological toll of cancer caregiving. Meaning-centered interventions (70,71) for cancer caregivers have mainly targeted those of patients who are at the palliative care phase and the outcomes studied are limited to self-reported and psychological distress (72). The present data suggest that finding meaning in caregiving may have the potential to at least partially offset the risk related aspects of caregiving, and may be relevant even in the earliest phases of the caregiving trajectory.

Given the complex interplay between risk and resilience processes, we also sought to determine which of these factors had the dominant effect in the analyses by including both domains simultaneously. In these analyses, loneliness stood out as the most robust correlate of CTRA gene expression, with none of the other psychological factors showing a distinct contribution to the variance in gene expression beyond its effects. Similar findings emerged when analyses additionally controlled for smoking and heavy alcohol consumption. One interpretation of this finding is that social support and meaning all ultimately exert their biological impacts via their effect on experiences of social isolation (i.e., loneliness). In other words, caregiving stress, lack of social support, and difficulty finding meaning each contributes to stress biology by hindering caregivers from achieving their desired social relationships (rather than via their own direct influences on caregiver biology). However, it is also possible that the social and meaning-related aspects of caregiving are correlated rather than mechanistically sequential, and that people who are successful in minimizing social isolation and achieving desired social relationships may also tend to effectively manage caregiving stress, mobilize social support, and find meaning in caregiving. Regardless of their causal sequence, our findings suggest that helping cancer caregivers achieve their desired social relationships to avoid loneliness may facilitate their physiological health.

Translational implications of the findings include a new intervention strategy to possibly mitigate caregivers’ health risks that incorporates ways for improving the quality of close social relationships with confidant and significant others to reduce emotional loneliness and for enhancing the general integration in the community to minimize social loneliness (73–75). Community- or oncology clinic-based volunteering may be a good example whereby family caregivers are involved in helping, prosocial behaviors for other patients while their patients are taken care of by other volunteers. This may increase a sense of emotional and social connections with other volunteer peers, mentors/mentees, and patients of their own and others’, as well as may mimic the salutary effects of respite from the demands and uncertainty of cancer caregiving to family members.

Several limitations of this study should be noted. First, although the current study used longitudinal CTRA data from family cancer caregivers and controlled for important covariates (e.g., age, gender/sex, ethnicity, BMI, and patients’ cancer stage), the results should nevertheless be interpreted with caution given the small sample size and correlational study design. The small sample size also limited any form of genome-wide discovery analyses, and thus we tested a single a priori “gene hypothesis” involving the CTRA (15). It is possible that caregiving is associated with other gene transcriptional correlates besides those documented here, and future research in larger samples will be required to identify such effects. The associations of psychological factors with CTRA may vary across the illness trajectory and psychological factors may relate to CTRA additively or synergistically (i.e., in interaction). Investigation of these important topics with larger sample size is warranted. Absence of a randomized non-caregiving control group also precludes causal claims, and this study’s focus on colorectal cancer caregivers in two US cities limits generalizability of our findings to other populations and other types of adverse events. The effect sizes of the CTRA differences we observed are small and their clinical significance is unknown. Long-term follow-up of these caregivers is necessary to adequately evaluate the biological cost of cancer caregiving. Future research should also consider examining physiological mediators of CTRA gene expression, such as sympathetic nervous system (SNS) activity (24,76) and parasympathetic activity that physiologically antagonize SNS effects via prosocial behaviors (40,42,77). Variability in the time of day of the blood collection in our sample is another limitation that introduces potential circadian effects, which should be investigated in a larger sample. Furthermore, our findings may help guide the development of new intervention strategies for family cancer caregivers, but cancer caregivers might also benefit from other types of existing cognitive-behavioral stress management (28), prosocial behaviors (38), and behavioral interventions, such as meditation, tai chi, and yoga (see 15 for review), which have proven impactful in other populations.

Despite these limitations, our findings shed new light on the psychological and molecular pathways through which caring for a cancer patient may affect a family member’s health. Caregiving-related CTRA biology appears to be most pronounced when caregivers are most attuned to the burdens and costs of caregiving (particularly social isolation and loneliness) and experience little meaning or purpose in their caregiving. These findings provide a rationale for the development of new intervention strategies that prioritize reductions in caregiver loneliness, and also minimize caregiving stress, strengthen social support, and promote the experience of meaning/purpose in caregiving, to ultimately enhance the health and resilience of familial caregivers of those with cancer.

Glossary

CTRA

Conserved Transcriptional Response to Adversity

BMI

body mass index