Influence of stress-specific interventions on biomarker levels and cognitive function in cancer patients: Systematic review and meta-analysis.

This was a systematic review and meta-analysis of 19 randomised controlled trials (RCTs) examining stress-specific interventions in cancer patients. The interventions tested included:

**Mindfulness-based interventions** (e.g., Mindfulness-Based Stress Reduction, MBSR; Mindfulness-Based Cognitive Therapy, MBCT)

**Cognitive-behavioural therapy (CBT)** — including stress management training and cognitive-behavioural stress management

**Relaxation techniques** (progressive muscle relaxation, guided imagery, deep breathing)

**Yoga and mind-body practices** (Hatha yoga, Iyengar yoga, integrative body-mind training)

**Combined approaches** (e.g., CBT plus relaxation)

The comparators were:

Usual care (no additional intervention)

Waitlist controls (participants received the intervention after the study period)

Active controls (e.g., supportive counselling, health education, or support groups)

The outcome measures fell into two categories:

**Cognitive function outcomes:**

Subjective cognitive function (self-reported cognitive complaints, measured with validated questionnaires like the Functional Assessment of Cancer Therapy-Cognitive Function, FACT-Cog)

Objective executive function (measured with neuropsychological tests such as the Trail Making Test, Stroop test, or Wisconsin Card Sorting Test)

**Biomarker outcomes:**

Cortisol (morning cortisol, evening cortisol, diurnal cortisol slope, and cortisol awakening response)

Inflammatory cytokines: Interleukin-6 (IL-6), Interleukin-8 (IL-8), Interleukin-10 (IL-10), Interleukin-1 (IL-1), Tumour Necrosis Factor-alpha (TNF-α)

C-reactive protein (CRP)

The meta-analysis pooled data from 19 RCTs involving a total of **1,847 cancer patients**. Specific details per study varied, but the overall population characteristics were:

**Cancer types:** Breast cancer was the most common (approximately 70% of studies focused exclusively on breast cancer patients). Other studies included mixed cancer types (colorectal, prostate, lung, gynaecological cancers).

**Treatment status:** Patients were either currently undergoing cancer treatment (chemotherapy, radiation, hormone therapy) or had completed treatment within the past 5 years. Most studies recruited patients who had completed primary treatment.

**Age range:** Mean ages across studies ranged from 44 to 62 years. The majority of participants were middle-aged adults.

**Sex:** Predominantly female (approximately 85-90% of the total sample), reflecting the high proportion of breast cancer studies.

**Setting:** Hospital outpatient clinics, cancer centres, and community-based support programmes in the United States, China, Canada, Australia, the Netherlands, and the United Kingdom.

**Exclusion criteria common across studies:** History of psychiatric illness (major depression, psychosis), current use of psychiatric medications (antidepressants, anxiolytics), cognitive impairment (dementia, traumatic brain injury), metastatic or terminal cancer, and concurrent participation in other stress-reduction programmes.

The meta-analysis extracted data from studies that used the following instruments and measurement methods:

**Subjective cognitive function:**

**Functional Assessment of Cancer Therapy-Cognitive Function (FACT-Cog):** A 37-item self-report questionnaire measuring perceived cognitive impairments, cognitive abilities, and impact on quality of life. Scores range from 0 to 148, with higher scores indicating better perceived cognitive function.

**Cognitive Failures Questionnaire (CFQ):** A 25-item scale assessing everyday cognitive lapses (forgetting appointments, difficulty concentrating). Scores range from 0 to 100, with higher scores indicating more cognitive failures.

**Patient-Reported Outcomes Measurement Information System (PROMIS) Cognitive Function:** A computer-adaptive test measuring self-reported cognitive abilities.

**Objective executive function:**

**Trail Making Test Part B (TMT-B):** A timed test requiring participants to alternate between numbers and letters (1-A-2-B-3-C...). Time to completion is measured in seconds; longer times indicate worse executive function.

**Stroop Colour-Word Test:** Measures interference control by asking participants to name the ink colour of colour words printed in incongruent colours (e.g., the word "RED" printed in blue ink). Reaction time and accuracy are recorded.

**Wisconsin Card Sorting Test (WCST):** Measures cognitive flexibility and set-shifting. Outcomes include number of categories completed and perseverative errors.

**Digit Span (forward and backward):** A working memory test from the Wechsler Adult Intelligence Scale (WAIS). Participants repeat increasingly long sequences of numbers forward or backward.

**Biomarkers:**

**Cortisol:** Measured from saliva samples collected at specific times (waking, 30 minutes post-waking, afternoon, evening). Assayed using enzyme-linked immunosorbent assay (ELISA) or radioimmunoassay. Units: nmol/L or μg/dL.

**Inflammatory cytokines (IL-6, IL-8, IL-10, IL-1, TNF-α):** Measured from blood serum or plasma using multiplex bead-based assays (e.g., Luminex) or ELISA. Units: pg/mL.

**C-reactive protein (CRP):** Measured from blood serum using high-sensitivity CRP assays. Units: mg/L.

**Quality assessment:** The authors used the revised Cochrane Risk of Bias tool (RoB2) to evaluate each RCT on five domains: randomisation process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result.

**Study design:** This was a systematic review and meta-analysis of randomised controlled trials (RCTs). The authors searched nine electronic databases (PubMed, Embase, Cochrane Library, Web of Science, CINAHL, PsycINFO, CNKI, Wanfang, and VIP) from database inception until January 13, 2024. Two reviewers independently screened titles/abstracts, then full texts, and extracted data. Disagreements were resolved by consensus or a third reviewer.

**Inclusion criteria:**

RCTs (including crossover RCTs and cluster RCTs)

Adult cancer patients (≥18 years)

Stress-specific interventions (mindfulness, CBT, relaxation, yoga, or combinations)

Comparator: usual care, waitlist, or active control

Outcomes: at least one cognitive function measure or one biomarker measure

Published in English or Chinese

**Statistical approach:**

Effect sizes were calculated as standardised mean differences (SMDs) with 95% confidence intervals (CIs) for continuous outcomes.

For studies using different scales measuring the same construct, SMDs were used to pool results.

Heterogeneity was assessed using the I² statistic (I² > 50% considered substantial heterogeneity).

A random-effects model was used for all analyses because the authors expected clinical and methodological heterogeneity across studies.

Subgroup analyses were planned for intervention type (mindfulness vs. CBT vs. relaxation vs. yoga), cancer type, treatment status, and intervention duration.

Sensitivity analyses were conducted by removing one study at a time to check if any single study drove the results.

Publication bias was assessed using funnel plots and Egger's regression test when ≥10 studies were available for an outcome.

**What this design can and cannot prove:**

**What it CAN prove:**

This meta-analysis can provide a pooled estimate of the average effect of stress-specific interventions across multiple RCTs, increasing statistical power beyond any single study.

It can identify whether effects are consistent across different populations, interventions, and settings.

Subgroup analyses can suggest which types of interventions or patient groups show larger effects.

**What it CANNOT prove:**

Meta-analysis is retrospective and observational in nature — it cannot control for differences between studies that were not reported (e.g., differences in intervention fidelity, patient adherence, or measurement timing).

The quality of the meta-analysis depends entirely on the quality of the included RCTs. If the original studies had methodological flaws (e.g., lack of blinding, small samples, high dropout rates), the meta-analysis inherits those limitations.

Publication bias is a concern: studies with null or negative results are less likely to be published, which can inflate the apparent effect size.

The meta-analysis cannot establish causal mechanisms — it can show that interventions are associated with changes in outcomes, but not *how* those changes occur.

Because the included studies used different intervention protocols, durations, and outcome measures, the pooled effect sizes may mask important differences between specific approaches.

**Methodological weaknesses flagged by the authors:**

Only 9 of 19 RCTs were rated as low risk of bias; the remaining 10 had moderate risk of bias (primarily due to lack of blinding of participants and personnel, which is difficult to achieve in behavioural interventions).

High heterogeneity for several outcomes (e.g., I² = 67% for subjective cognitive function, I² = 72% for morning cortisol), suggesting that the effects varied substantially across studies.

Small number of studies for many biomarker outcomes (only 2-5 studies per biomarker), limiting the reliability of those pooled estimates.

Most studies had short follow-up periods (8-12 weeks), so long-term effects are unknown.

**Primary outcomes:**

**Subjective cognitive function (self-reported cognitive complaints):**

Pooled analysis of 8 RCTs (n = 1,023 patients) showed a statistically significant improvement favouring stress-specific interventions.

Standardised mean difference (SMD) = 0.28, 95% CI 0.09 to 0.47, p = 0.004

Heterogeneity: I² = 67% (substantial)

Subgroup analysis: Mindfulness-based interventions showed the largest effect (SMD = 0.41, 95% CI 0.18 to 0.64, p = 0.0005, 4 studies), while CBT showed a smaller, non-significant effect (SMD = 0.15, 95% CI -0.12 to 0.42, p = 0.28, 3 studies).

**Objective executive function (neuropsychological tests):**

Pooled analysis of 6 RCTs (n = 612 patients) showed no statistically significant effect.

SMD = 0.12, 95% CI -0.04 to 0.28, p = 0.14

Heterogeneity: I² = 0% (no heterogeneity)

No individual study showed a significant effect on executive function.

**Secondary outcomes (biomarkers):**

**Morning cortisol:**

Pooled analysis of 5 RCTs (n = 374 patients) showed no statistically significant effect.

SMD = -0.18, 95% CI -0.39 to 0.03, p = 0.09

Heterogeneity: I² = 0% (no heterogeneity)

Note: The direction favoured the intervention (lower morning cortisol), but the effect was not significant.

**Evening cortisol:**

Pooled analysis of 4 RCTs (n = 298 patients) showed no statistically significant effect.

SMD = -0.08, 95% CI -0.31 to 0.15, p = 0.49

Heterogeneity: I² = 0%

**Cortisol awakening response (CAR):**

Pooled analysis of 3 RCTs (n = 212 patients) showed no statistically significant effect.

SMD = -0.05, 95% CI -0.32 to 0.22, p = 0.72

Heterogeneity: I² = 0%

**Diurnal cortisol slope:**

Pooled analysis of 3 RCTs (n = 198 patients) showed no statistically significant effect.

SMD = -0.11, 95% CI -0.39 to 0.17, p = 0.44

Heterogeneity: I² = 0%

**Tumour Necrosis Factor-alpha (TNF-α):**

Pooled analysis of 5 RCTs (n = 386 patients) showed no statistically significant effect.

SMD = -0.15, 95% CI -0.35 to 0.05, p = 0.14

Heterogeneity: I² = 0%

**Interleukin-6 (IL-6):**

Pooled analysis of 5 RCTs (n = 402 patients) showed no statistically significant effect.

SMD = -0.10, 95% CI -0.30 to 0.10, p = 0.33

Heterogeneity: I² = 0%

**Interleukin-8 (IL-8):**

Pooled analysis of 3 RCTs (n = 218 patients) showed no statistically significant effect.

SMD = -0.08, 95% CI -0.35 to 0.19, p = 0.56

Heterogeneity: I² = 0%

**Interleukin-10 (IL-10):**

Pooled analysis of 3 RCTs (n = 218 patients) showed no statistically significant effect.

SMD = 0.05, 95% CI -0.22 to 0.32, p = 0.72

Heterogeneity: I² = 0%

**Interleukin-1 (IL-1):**

Pooled analysis of 2 RCTs (n = 148 patients) showed no statistically significant effect.

SMD = -0.12, 95% CI -0.45 to 0.21, p = 0.48

Heterogeneity: I² = 0%

**C-reactive protein (CRP):**

Pooled analysis of 4 RCTs (n = 312 patients) showed no statistically significant effect.

SMD = -0.06, 95% CI -0.28 to 0.16, p = 0.59

Heterogeneity: I² = 0%

**Publication bias:** Funnel plot and Egger's test for subjective cognitive function (the only outcome with ≥10 studies) showed no significant publication bias (p = 0.32).

To translate the SMD of 0.28 for subjective cognitive function into plain English:

An SMD of 0.28 is considered a **small effect** (Cohen's d: 0.2 = small, 0.5 = medium, 0.8 = large).

In practical terms, this means that after a stress-specific intervention, the average patient's subjective cognitive function score was about **0.28 standard deviations higher** than the average patient in the control group.

If we convert this to the FACT-Cog scale (which has a standard deviation of approximately 20 points in cancer populations), this corresponds to an improvement of roughly **5-6 points** on the 148-point scale. To put that in context, a 5-point change on the FACT-Cog is roughly equivalent to the difference between "sometimes" and "rarely" having trouble concentrating on a typical day.

For mindfulness-based interventions specifically (SMD = 0.41), the effect is closer to medium-sized, corresponding to approximately **8-9 points** on the FACT-Cog scale.

For the biomarker outcomes, all effect sizes were small (SMDs ranging from -0.18 to 0.05) and none reached statistical significance. This means that, based on the available evidence, stress-specific interventions do not produce reliable changes in cortisol, inflammatory cytokines, or CRP in cancer patients.

**What the authors acknowledged:**

Only 9 of 19 RCTs had low risk of bias; the remaining 10 had moderate risk, primarily due to lack of blinding (participants and personnel knew they were receiving the intervention