Effect of serum concentrations of IL-6 and TNF-α on brain structure in anorexia nervosa: a combined cross-sectional and longitudinal study

Abstract

Previous studies of brain structure in anorexia nervosa (AN) have reported reduced gray matter in underweight patients, which largely normalizes upon weight gain. One underlying biological mechanism may be glial cell alterations related to low-grade inflammation. Here, we investigated relationships between brain structure as measured by magnetic resonance imaging and serum concentrations of two pro-inflammatory cytokines (interleukin-6 and tumor necrosis factor alpha) cross-sectionally in 82 underweight adolescent and young adult female patients (mean age 16.8 years; 59 of whom were observed longitudinally after short-term weight restoration; mean duration 2.8 months), 20 individuals long-term weight-recovered from AN (mean age 22.7 years) and 105 healthy control (HC) participants (mean age 17.2 years). We measured cortical thickness, subcortical volumes and local gyrification index, a measure of cortical folding. In contrast to most previous studies of cytokine concentrations in AN, we found no cross-sectional group differences (interleukin-6: p = 0.193, tumor necrosis factor alpha: p = 0.057) or longitudinal changes following weight restoration (interleukin-6: p = 0.201, tumor necrosis factor alpha: p = 0.772). As expected, widespread gray matter reductions (cortical thickness, subcortical volumes, cortical folding) were observed in underweight patients with AN compared to HC. However, we found no evidence of associations between cytokine concentrations and structural brain measures in any participant group. Furthermore, longitudinal changes in cytokine concentrations were unrelated to changes in gray matter. In conclusion, we did not identify any association between (sub-)inflammatory processes and structural brain changes in AN. Future studies are needed to elucidate which other factors besides nutritional status may contribute to brain morphological alterations.

Fig. 1. Serum IL-6 concentrations.

Violin plots showing the median, upper and lower quartile, outliers (depicted as circles, values deviating more than 1.5 times the interquartile range from the upper or lower quartile) and the kernel probability density of the data. Serum IL-6 concentrations (pg/ml): HC: Median = 0.62, interquartile range = 0.70, range = <0.03–10.44; acAN-T1: Median = 0.51, interquartile range = 0.61, range = <0.03 to 9.82; acAN-T2: Median = 0.52, interquartile range = 0.75, range = 0.08–7.13; recAN: Median = 0.63, interquartile range = 0.63, range = 0.27–6.45. The figure was created with JASP [65]. acAN-TP1 acutely underweight participants with AN at timepoint 1 (admission), acAN-TP2 participants with AN at timepoint 2 (after short-term weight restoration), HC healthy control participants, recAN individuals long-term weight-recovered from AN.

Fig. 2. Serum TNF-α concentrations.

Violin plots showing the median, upper and lower quartile, outliers (depicted as circles, values deviating more than 1.5 times the interquartile range from the upper or lower quartile) and the kernel probability density of the data. Serum TNF-α concentrations (pg/ml): HC: Median = 0.69, interquartile range = 0.23, range = 0.35–2.18; acAN-T1: Median = 0.63, interquartile range = 0.22, range = 0.29–2.75; acAN-T2: Median = 0.62, interquartile range = 0.28, range = 0.30–1.79; recAN: Median = 0.56, interquartile range = 0.32, range = 0.33–1.30. The figure was created with JASP [65]. acAN-T1 acutely underweight participants with AN at timepoint 1 (admission), acAN-T2 participants with AN at timepoint 2 (after short-term weight restoration), HC healthy control participants, recAN individuals long-term weight-recovered from AN.

Fig. 3. No significant associations between longitudinal changes in cortical structure (acAN-TP2-acAN-TP1) and changes in serum concentrations of the IL-6 cytokine.

Left: Associations with CT changes. Right: Associations with lGI changes. Uncorrected statistical maps (p < 0.05) plotted on the inflated surface of the standard average subject and displaying regions in which differences in CT or lGI between acAN-TP2 and acAN-TP1 were associated with changes in IL-6. The color scale shows p values expressed as −log₁₀(p). Warm colors indicate a positive correlation between changes in IL-6 concentrations and CT or lGI. After FDR-correcting for multiple comparisons (q < 0.05) across each hemisphere and Bonferroni correcting across cytokine type, these associations were not significant. LH left hemisphere, RH right hemisphere. Colored outlines correspond to anatomical labels of the Desikan-Killiany atlas [44].

Fig. 4. No significant associations between longitudinal changes in cortical structure (acAN-TP2-acAN-TP1) and changes in serum concentrations of the TNF-α cytokine.

Left: Associations with CT changes. Right: Associations with lGI changes. Uncorrected statistical maps (p < 0.05) plotted on the inflated surface of the standard average subject and displaying regions in which differences in CT or lGI between acAN-TP2 and acAN-TP1 were associated with changes in TNF-α. The color scale shows p values expressed as −log₁₀(p). Warm colors indicate a positive correlation between changes in TNF-α and CT or lGI. After FDR-correcting for multiple comparisons (q < 0.05) across each hemisphere and Bonferroni correcting across cytokine type, these associations were not significant. LH left hemisphere, RH right hemisphere. Colored outlines correspond to anatomical labels of the Desikan-Killiany atlas [44].