Depression explains the association between pain intensity and pain interference among adults with neurofibromatosis

Abstract

Introduction

Neurofibromatoses (NFs; NF1, NF2 and Schwannomatosis) are incurable genetic syndromes characterized by nerve sheath tumors and often accompanied by substantial emotional distress (e.g., depression and anxiety). Pain is also common but understudied in adults with NF and interferes with daily living. In other medical populations, depression and anxiety have a strong association with pain interference. However, research has not explored the relationship of depression and anxiety to pain interference among adults with NF experiencing pain. The aim of this study was to test the hypothesis that depression and anxiety will mediate the association between pain intensity and pain interference among geographically diverse adults with NF who endorse pain.

Methods

We used baseline data from an RCT of a mind–body intervention aimed at improving quality of life in adults with NF. Participants (N = 214) who endorsed pain completed measures of demographics, clinical characteristics, baseline pain intensity, pain interference, depression, and anxiety. We constructed a multiple mediation model in R using the lavaan package to test our hypothesis.

Results

Preliminary analyses showed differences in pain interference by NF diagnostic subtype (F(2, 206) = 6.82, p = 001). In a model that controlled for NF diagnostic subtype, we found that depression β = .07, p = .017), but not anxiety (β = −.003, p = .878), partially mediated the association between pain intensity and pain interference.

Conclusion

Improving depression has the potential to decrease pain interference among people with NF who experience pain.

Trial registration

Clinicaltrials.gov Registration #: NCT03406208

Introduction

Neurofibromatoses (NF) are heterogeneous, incurable genetic syndromes (NF1, NF2 and Schwannomatosis) characterized by nerve sheath tumors [1]. These tumors and other NF complications often cause intense pain [1-4]. While pain is a core feature of Schwannomatosis (SCHWN) [5], pain is prevalent across NF diagnostic subtypes, including NF1 and 2 [6, 7], and significantly interferes with daily functioning. Available treatment options for adults with NF primarily consist of surgery or medication (e.g., opioids) [3, 8], both of which have considerable limitations including side effects, significant risks, and limited clinical benefit [9-12]. A survey of 255 adults with NF found that almost half (47%) of participants that underwent surgery in the past year reported post-surgical complications, and 85% of those receiving opioids indicated the medication had “little to no effectiveness” [9]. Identifying modifiable factors associated with pain interference among people with NF is therefore a high priority and can inform psychosocial interventions.

Mounting evidence supports biopsychosocial models of care for successfully addressing pain across chronic illnesses including NF [13-18]. These models can inform approaches to pain treatment that account for links between pain and emotional distress [19]. Evidence suggests emotional distress (e.g., depression, anxiety) is common among adult individuals with NF [18, 20, 21], and that biological underpinnings of emotional distress (e.g., heart rate variability) are associated with pain interference in daily functioning [22]. While emotional distress has been found to worsen pain interference in other medical populations [23, 24], the role of emotional distress in pain interference among adults with NF experiencing pain remains largely unknown.

Here, we tested baseline associations among pain intensity, depression, anxiety, and pain interference among geographically diverse adults with NF who enrolled in a mind–body symptoms and stress management clinical trial and endorsed pain. We hypothesized that depression and anxiety, tested simultaneously within one model, would both mediate the association between pain intensity and pain interference in this sample.

Method

Participants and procedures

We used baseline data from a virtual efficacy trial of a mind–body program aimed at improving quality of life among adults with NF versus an attention placebo control. Recruitment occurred between May 2017 and February 2021. Out of the 228 adults enrolled (158 with NF1, 31 with NF2, 30 with SCHWN), we selected 214 who endorsed pain intensity scores higher than 0 over the past week on the Graded Chronic Pain Scale (154 with NF1, 31 with NF2, 29 with SCH). The published clinical trial protocol describes the full methodology, including recruitment strategy and inclusionary and exclusionary criteria [25]. Briefly, participants were recruited from the United States and internationally, primarily from the Children’s Tumor Foundation, through a monthly electronic mailing list. Eligible participants needed to report stress and difficulties coping with NF symptoms, endorse a score of 6 or higher on the Perceived Stress Scale-6, be able to understand English at a 6^th grade level, be on stable medications, free of substance use disorders, without major surgery scheduled for the next 6 months, and able to participate in live video groups with geographically diverse participants. All participants signed informed consent prior to enrollment. Enrolled participants completed baseline questionnaires through REDCap prior to randomization [26]. For the present study, we were interested in pain intensity, pain interference, and symptoms of emotional distress, specifically.

Measures

Demographic and clinical variables

Participants reported their age, sex, race, ethnicity, education, marital status, and NF diagnostic subtype (NF1, NF2, SCHWN).

Depression

The 9-item Patient Health Questionnaire (PHQ-9) [27] measures self-reported frequency of DSM-5 depressive symptoms within the past two weeks on a 4-point Likert scale from 0 (“Not at All”) to 3 (“Nearly Every Day”) [28]. Higher scores represent greater depressive symptoms (sample Chronbach’s α = 0.81).

Anxiety

The Generalized Anxiety Disorder-7 (GAD-7) [29] is a 7-item self-report scale that measures the frequency of DSM-5 generalized anxiety disorder symptoms within the past two weeks on a 4-point Likert scale from 0 (“Not at All”) to 3 (“Nearly Every Day”) [28]. Higher scores indicate anxiety symptoms (sample α = 0.91).

Pain intensity

The Graded Chronic Pain Scale (GCPS) [30] measures current pain along with worst pain and average pain over the past 6 months on 11-point numerical rating scales (“0 = no pain” to “10 = pain as bad as it could be”). As done in other published pain studies [31, 32], we adapted the GCPS to measure worst pain and average pain within the past week to ensure that we captured pain symptoms proximal to other measured variables (sample α = 0.92).

Pain interference

The 8-item PROMIS v1.0—Pain Interference—Short Form 8a (PROMIS—PI8) [33] measures self-reported consequences of pain on relevant aspects of one’s life over the past 7 days (e.g., household chores, social activities, occupational tasks). Participants respond to items using a 5-point Likert scale from 1 (“Not at All”) to 5 (“Very Much”). Items are summed to create a total raw score, which is converted to a T-score (i.e., standardized score; M = 50, SD = 10) (sample α = 0.98).

Statistical analysis

We excluded 14 participants who endorsed scores of 0 on the GCPS pain intensity subscale (12 with NF1, 1 with NF2, and 1 with SCHWN). Next, we ran bivariate correlations and one-way analyses of variance (ANOVAs) to test whether demographic and clinical characteristics (e.g., age, sex, diagnostic subtype, race, ethnicity) were associated with the outcome variable (pain interference). Any variables found to be associated with the outcome variable were included as co-variates in the multiple mediation analyses (see preliminary analyses). All variables were standardized as z scores prior to analyses to account for differences in scaling. Guided by established procedures [34, 35], we conducted multiple mediation in R and RStudio 3.6.1 using a structural equation modeling framework with the lavaan package [36-39]. Multiple mediation analyses allowed us to test the effect of one proposed mediator while accounting for effect of the other with reduced likelihood of parameter bias due to omitted variables [35]. This was particularly important for our study given strong correlations between depression and anxiety in the general population and in the present sample (r = 0.68, p < 0.001). Per published recommendations for mediation analysis [40, 41], we tested the assumption that data were missing completely at random (MCAR) in R using the naniar package [42]. We handled missing data in our analysis using full-information maximum likelihood (FIML) estimation, a recommended approach [34], which utilizes all observed variables for each case. FIML is more flexible than other approaches to missing data, allowing data to be either MCAR or missing at random (MAR).

To test our hypothesis, we estimated the effect of pain intensity on depression (path a1) and anxiety (path a2), the effects of depression and anxiety on pain interference (paths b1 and b2, respectively), the indirect effect of pain intensity on pain interference through both depression (path a1 × b1) and anxiety (path a2 × b2), and the direct effects of pain intensity (path c′) and any covariates on interference, accounting for the effects of the proposed mediators. We assessed the magnitude of effects using standardized beta coefficients. We evaluated the reliability of mediation effects using percentile (nonparametric) 95% confidence intervals (CIs) generated by a bootstrapping procedure with a resample rate of 10,000 [35, 43, 44]. We evaluated model fit for the multiple mediation model using recommended fit indices: the chi-square exact test of model fit, the Root Mean Square Error of Approximation (RMSEA), the Standardized Root Mean Square Residual (SRMR), and the Comparative Fit Index (CFI) [34]. We report variance in pain interference explained by the focal predictor (pain intensity) and by each mediation path (depression and anxiety).

Results

Preliminary analyses

Four participants with insufficient data were excluded via FIML procedures, resulting in a sample of 210 participants for the primary analysis (see Table 1 for participant characteristics): 154 with NF1, 31 with NF2, and 29 with SCHWN. Descriptive statistics and bivariate correlations among study variables are reported in Table 2. One-way ANOVA revealed that mean scores on pain interference differed significantly by NF diagnostic subtype (F(2, 206) = 6.82, p = 001), with individuals diagnosed with SCHWN (M = 28.79, SD = 10.15) having higher mean scores than those with NF1 (M = 21.70, SD = 9.87) or NF2 (M = 21.06, SD = 8.99). As such, we entered dummy coded diagnosis variables for NF1 (NF1 = 1, NF2 = 0, SCHWN = 0) and NF2 (NF2 = 1, NF1 = 0, SCHWN = 0) as covariates/direct paths in the multiple mediation model (SCHWN was the reference variable). Pain interference was not associated with any other demographic variables (ps > 0.067).

Table 1.

Participant characteristics

Participant characteristics (N = 214)	Participant, No. (%)
Diagnosis
NF1	154 (72.0%)
NF2	31 (14.5%)
Schwannomatosis	29 (13.6%)
Gender
Male	57 (26.6%)
Female	157 (73.4%)
Age	Mean = 42.96 (SD = 14.40)
Marital status
Married	91 (42.5%)
Living with someone in a committed relationship	15 (7.0%)
Single	86 (40.2%)
Separated	2 (0.9%)
Divorced	15 (7.0%)
Widowed	3 (1.4%)
Chose not to answer	2 (0.9%)
Education level
Less than high school	16 (7.5%)
Completed high school	22 (10.3%)
Some college or associate degree	56 (26.2%)
Completed 4 years of college	47 (22.0%)
Graduated or professional degree	73 (34.1%)
Race
White	180 (84.1%)
Black/African American	8 (3.7%)
Asian	7 (3.3%)
American Indian/Alaskan Native	1 (0.5%)
More than one race	12 (5.6%)
Chose not to answer	6 (2.8%)
Ethnicity
Hispanic or Latino/Latina	12 (5.6%)
Not Hispanic or Latino/Latina	198 (92.5%)
Chose not to answer	4 (1.9%)

Table 2.

Means, standard deviations, and correlations among study

	1	2	3	4
1. Depression	–
2. Anxiety	.68*	–
3. Pain intensity	.40*	.29*	–
4. Pain interference	.44*	.30*	.74*	–
Descriptives
Full sample (N = 214)
Scale	0–27	0–21	0–100	8–40
M	10.58	9.30	51.65	22.59
SD	5.55	5.70	24.38	10.06
NF1 (n = 154)
M	10.84	9.54	51.62	21.70
SD	5.37	5.59	24.39	9.87
NF2 (n = 31)
M	8.43	7.85	43.23	21.06
SD	4.88	5.57	20.27	8.99
SCHWN (n = 29)
M	11.41	9.55	60.80	28.79
SD	6.70	6.31	25.80	10.15

Levels of depression anxiety were similar between the initial sample (N = 228, PHQ-9: M = 10.70, SD = 5.84; GAD-7: M = 9.44, SD = 5.78) and the final sample (N = 214, PHQ-9: M = 10.58, SD = 5.55; GAD-7: M = 9.30, SD = 5.70), containing only participants who endorsed pain intensity scores above 0

Depression = PHQ-9. Anxiety = GAD-7. Pain Intensity = GCPS Pain Intensity subscale. Pain Interference = PROMIS – PI8

^*p < .01

Primary analysis

Mediation through depression and anxiety

The chi-square test of exact model fit suggested that the model-implied variance–covariance matrix adequately reproduced the data, χ²(4) = 2.76, p = 0.598. Additional indexes also met the criteria for acceptable model fit (RMSEA = 0.000, SRMR = 0.031, CFI = 1.000) [45]. Multiple mediation results showed that greater pain intensity was associated with greater pain interference (path c: β = 0.73, SE = 0.04, 95% CI = [0.65, 0.81], p < 0.001), depression (path a1: β = 0.40, SE = 0.07, 95% CI = [0.27, 0.53], p < 0.001) and anxiety (path a2; β = 0.29, SE = 0.07, 95% CI = [0.15, 0.42], p < 0.001). However, only greater depression (path b1; β = 0.18, SE = 0.07, 95% CI = [0.04, 0.31], p = 0.009), not anxiety (path b2; β = −0.01, SE = 0.07, 95% CI = [− 0.15, 0.12], p = 0.873), was associated with greater pain interference while controlling for the effects of diagnostic subtype. Regarding covariates, a diagnosis of NF1 was associated with significantly less pain intensity than SCHWN (β = − 0.43, SE = 0.11, 95% CI = [− 0.65, −0.22], p < 0.001) while a diagnosis of NF2 was not (β = − 0.20, SE = 0.15, 95% CI = [−0.49, 0.09] p = 0.178).

The indirect effect of pain intensity on pain interference was significant through depression (path a1*b1: β = 0.07, SE = 0.03, 95% CI = [0.02, 0.14], p = 0.017) but not anxiety (path a2*b2: β = −0.003, SE = 0.02, 95% CI = [−0.05, 0.04], p = 0.878). The direct effects of pain intensity (path c′: β = 0.66, SE = 0.05, 95% CI = [0.57, 0.75], p < 0.001) remained significant after adding our proposed mediators to the model, suggesting the mediation effects for depression were partial (Fig. 1). In this model, pain intensity and depression explained 60% and 16% of the variance in pain interference, respectively.

Fig. 1.

Mediation model testing the indirect effects of pain intensity on pain interference via depression and anxiety, controlling for diagnostic subtype. Bootstrapped standard errors are reported in parentheses. Bootstrapped 95% confidence intervals are reported in brackets. *p < .05

Discussion

We tested associations among pain intensity, depression, anxiety, and pain interference using baseline data from geographically diverse adults with pain who enrolled in an efficacy trial of a virtual mind–body program to enhance quality of life. We showed that higher pain intensity was associated with higher depression, anxiety and pain interference. However, only depression, not anxiety, partially explained the association of higher pain intensity to higher pain interference.

Our findings suggest that, among adults with NF who endorse pain, improving depression may be associated with reductions in pain interference and help individuals engage in activities of daily living. Pain interference was positively correlated with all PHQ-9 items except for suicidality (rs = 0.15 = 0.42, ps = 0.000–0.032). Thus, most evidence-based approaches shown to target symptoms of depression (e.g., cognitive behavioral therapy, mindfulness and mind–body based interventions) may reduce pain interference in this population [18, 46, 47]. Our group is iteratively developing a virtual mind–body intervention teaching mindfulness, relaxation, and pain coping skills to adults with NF to enhance overall quality of life [48]. This intervention targets emotional distress (including depression), and, thus, may be efficacious in reducing pain interference. Our prior work also suggests that mind–body interventions may decrease anxiety, depression, and pain interference among adolescents [48]. Because NF is often diagnosed in childhood, and psychosocial factors are associated with pain interference among youth [4], early mind–body skills training may help reduce pain interference throughout the lifespan.

There are two potential explanations for why anxiety was not a mediator of the association between pain and pain interference in our sample. First, depression and anxiety are highly comorbid [49, 50], and depression, compared to anxiety, was more strongly correlated with pain interference in our sample. We chose to test the potential role of depression and anxiety within one model to better account for what is naturally occurring among the general NF population. As a result, our model may have underestimated the effects of anxiety on pain interference. Second, we used the GAD-7 to measure anxiety, which captures general anxiety and worry. It is possible that another scale specifically targeting pain anxiety, such as the pain anxiety symptoms scale [51], would have yielded stronger effects.

Of note, individuals with SCHWN had significantly greater pain interference (M = 28.79) than those with NF1 (M = 21.70) or NF2 (M = 21.06). This is consistent with prior research demonstrating that pain is common and challenging to address in this population [5]. Interestingly, rates of pain (i.e., scores > 0 on the GCPS pain intensity subscale) were high in our sample. Out of 228 individuals enrolled in the mind–body trial, 214 (94%) endorsed at least some pain; 154 out of 166 patients with NF1 (93%), 31 out of 32 patients with NF2 (97%) and 29 out of 30 (97%, with SCHWN). This highlights the importance of assessing and addressing pain for all adults with NF. These results should be interpreted with caution, however, as we only enrolled individuals who endorsed stress (PSS-4 > 6), and stress and pain are highly co-morbid [52, 53]. Further, it is unclear if the pain reported by these participants is related to NF, comes from other sources, or is a combination of both. Nonetheless, because we controlled for NF diagnostic subtype in our mediation analysis, results suggest that depression is important in the association of pain and pain interference for adults with NF, regardless of diagnosis.

Limitations of this study should be noted. First, the cross-sectional design prevents us from establishing causal relationships between the examined constructs. It is likely that the relationship among variables in bidirectional. Second, despite accounting for NF diagnostic subtype in our mediation analyses, smaller group sizes for NF2 (n = 31) and SCHWN (n = 29) may limit the generalizability of our findings across NF diagnostic subtypes. Future studies with longitudinal designs and larger samples across diagnostic subtypes should help establish causal links among pain intensity, emotional factors (e.g., depression), and pain interference or other markers of quality of life. Finally, we recruited participants as part of a mind–body clinical trial for stress and symptom management and required participants to endorse a threshold level of stress in order to participate. As such, results cannot generalize to the entire adult population with NF.

Strengths of this study should also be noted. First, this study explores associations between pain and emotional functioning among adults across the NF-diagnostic spectrum while many other studies focus on a narrower range of NF patients (e.g., selecting for one NF diagnostic subtype) [4, 5]. Second, we enrolled geographically diverse adults from across US and internationally. Further, while other studies demonstrate the debilitating effects of pain and its interference in daily life among individuals with NF [4, 5, 21], the current study suggests a potential mechanism by which it could be curbed, which can be targeted within biopsychosocial models of care to enhance the well-being of NF patients despite pain.

Conclusion

Improving depression holds the potential to prevent pain from significantly interfering in the daily lives of patients with NF. Upcoming data from the RCT of a mind–body intervention, in which participants in the current study were enrolled, may provide important information testing this model longitudinally.