Introduction
Hashimoto’s thyroiditis (HT) is the most common autoimmune thyroid condition and the overall cause of hypothyroidism in the Western world, disproportionately affecting Caucasian females over men and other ethnic groups [1]. HT is a complex disease with multiple etiologic factors, including environmental exposures, drug use, pregnancy, nutritional intake, and infectious diseases [2]. Family and twin-based studies have revealed various genetic susceptibilities primarily related to variations in an individual’s human leukocyte antigen (HLA) genotype as well as variations in numerous cytokines and the vitamin D receptor [2]. Given the complexity of HT, with numerous genetic contributors and our emerging understanding of additional environmental mediators, further research in therapies that can positively modify known environmental factors and mitigate risk for genetically susceptible individuals is warranted.
Currently, for individuals diagnosed with HT, there are few, if any, efficacious treatments outside of thyroid hormone replacement. Winther et al. showed that, in a cohort of 78 consecutive individuals newly diagnosed with HT, baseline markers of quality of life, as measured by the 36-Item Short Form Health Survey (SF-36), were significantly lower than normative healthy controls [3]. Despite slight improvements in thyroid-specific and mental-health-specific quality of life, individuals with HT persisted with overall lower quality of life as compared to healthy controls even after six months of treatment with levothyroxine therapy [3]. Additionally, even after normalizing thyroid function via hormonal replacement, many individuals with HT persist with numerous symptoms, such as chronic fatigue, dry skin, hair loss, chronic irritability, and nervousness, impairing quality of life [4].
To address the need for additional therapeutic options targeted at improving quality of life and symptom burden in individuals with HT, the objective of this study was to determine the efficacy of a multi-week diet and lifestyle intervention implemented by a physician, a team of nutritional therapy practitioners (NTPs), and health coaches. The study authors hypothesized that the multi-dimensional intervention would improve the participant's HRQL as well as decrease the participant's clinical symptom burden. The study authors additionally hypothesized that the intervention would improve thyroid function as measured by a decrease in TSH and increases in free and total T3 and T4. It was speculated that some individuals would require less thyroid replacement medication after the 10-week intervention. Finally, the study authors sought to explore the effect of the 10-week intervention on inflammation and immune activity as measured by high-sensitivity C-reactive protein (hs-CRP), white blood cell (WBC) count, differential cell counts, and thyroid antibodies, including thyroid peroxidase (TPO) antibodies and anti-thyroglobulin antibodies (TGA).
There have been numerous clinical trials evaluating the use of dietary interventions for a variety of autoimmune diseases, including inflammatory bowel disease (IBD), multiple sclerosis, psoriasis, celiac disease, autoimmune thyroiditis, and rheumatoid arthritis [5-11]. A recent 2017 review assessed the role of iodine, selenium, vitamin D, and gluten on the management of patients with HT [10]. The authors concluded that the role of a gluten-free diet may be of benefit for patients with HT independent of a comorbid diagnosis of celiac disease. Despite the potential benefit from this dietary elimination, they speculated, however, that quality of life could be negatively impacted given the restrictive nature of the gluten-free diet.
Outside of our current understanding regarding the importance of certain key nutrients for the optimal functioning of the thyroid gland, namely iodine, selenium, zinc, iron, B12, and lipid soluble vitamins, including A, E, D, and K, there are no specific dietary guidelines or recommendations for individuals with autoimmune thyroid disease. While some have speculated that over-consumption of dietary goitrogens could negatively impact thyroid functioning, there is an absence of rigorous trials suggesting the negative effects of such foods when consumed in normal proportions.
In seeking to identify a dietary template and feasible lifestyle intervention that could positively improve HRQL and symptom burden in individuals with HT, five criteria were outlined: (1) Consumption of foods high in micronutrients containing, but not limited to, the aforementioned nutrients essential for thyroid functioning; (2) Elimination of foods with low nutritional value (sugar-sweetened beverages, ultra-processed foods, etc.) and foods that could result in an aberrant immune response via dysregulated antigen presentation or detrimentally affect both the gut microbiome and the integrity of the gastrointestinal barrier; (3) Implementation of dietary changes in phases, utilizing education and support from health coaches, NTPs, and a physician to improve dietary adherence; (4) Facilitation of the dietary and lifestyle intervention as part of an online community allowing participants to engage with other study members; (5) Prior clinical evidence for the efficacy of the intervention for a specific autoimmune disease.
In light of these five criteria, the Autoimmune Protocol (AIP), as implemented within the “SAD (Standard American Diet) to AIP in SIX” online, community-based health coaching program was the dietary and lifestyle program identified as most likely to result in participant adherence and symptom improvement. AIP, as implemented by the “SAD to AIP in SIX” program, was previously studied in individuals with inflammatory bowel disease (IBD) and was shown to be able to induce remission and improve symptoms in over 70% of participants [5].
In terms of its dietary composition, AIP is a modification of the Paleolithic diet that begins with an elimination of specific foods, dietary additives, emulsifiers, and western dietary patterns that have been implicated in disrupting the flora of the gastrointestinal microbiome as well as the intestinal barrier, leading to dysregulated antigen presentation and the development of autoimmunity [12-17]. In addition to the eliminated foods, AIP additionally promotes the consumption of nutrient-dense whole foods such as vegetables, fruits, mono and polyunsaturated fatty acids, tubers, wild game, poultry, organ meats, and non-processed meats.
Materials and methods
Study design and measures
Prior to enrolling in the trial, advertising for the study was completed across various social media outlets and local practitioners treating patients with HT. Upon receiving correspondence from interested participants, communication was initiated by study investigators to assess the participants’ ability to enroll in the intervention. A total of 456 subjects were screened with inclusion and exclusion criteria, resulting in 17 subjects eligible for inclusion in the trial. Inclusion criteria consisted of English-speaking male or females, 20-45 years of age, with a diagnosis of HT and a body mass index (BMI) between 18.4 and 29.9. Exclusion criteria consisted of individuals outside the listed age or BMI criteria, no definitive diagnosis of HT, prior experience with AIP for >30 days, pregnant, breastfeeding, six months postpartum, presence of other comorbidities, including hypertension, diabetes, heart disease, heart failure, liver failure, chronic or end-stage kidney disease, use of medications outside of Food and Drug Administration (FDA)-approved thyroid replacement medications, or an individual being unable to complete a two-week washout period before the start of the trial. In selecting the inclusion and exclusion criteria, the study authors sought to identify normal or overweight (non-obese) premenopausal women to minimize the influence of hormonal variations between pre and post-menopausal women as well as to minimize the likelihood for rapid weight loss in obese individuals. The study authors sought to minimize the risk of adverse effects, complications, and variations in thyroid function secondary to other disease processes by excluding individuals with chronic organ disease/organ failure as well as pregnant or breastfeeding women and women in the early post-partum period.
The two-week washout period consisted of a screening visit prior to the initiation of the formal dietary intervention where subjects signed informed consent, provided demographic information, completed study questionnaires, including a validated quality of life survey, the 36-Item Short Form Health Survey (SF-36), the Cleveland Clinic Center for Functional Medicine’s Medical Symptom Questionnaire (MSQ), and the National Institutes of Health’s (NIH's) food frequency questionnaire (FFQ) [18-19]. Participants provided fasting blood samples, including complete blood cell count with differential (CBC), complete metabolic profile (CMP), thyroid stimulating hormone (TSH), free T4, free T3, total T4, total T3, reverse T3, thyroid peroxidase (TPO) antibodies, anti-thyroglobulin antibodies (TGA), 25-hydroxycholecalciferol, and high sensitivity C-reactive protein ( hs-CRP). In addition, Genova Diagnostics Laboratory supplied organic acid tests (NutrEval FMV™, Genova Diagnostics Laboratory, NC, USA) as well as comprehensive stool analysis (GI Effects™, Genova Diagnostics Laboratory, NC, USA) for participants to complete during the washout period.
After completion of the washout period, participants began a 10-week online dietary and lifestyle intervention, which consisted of a six-week process of food elimination, the addition of nutrient-dense foods, and a focus on lifestyle modifications, followed by a four-week maintenance phase (during which no food group reintroductions were allowed), using the “SAD (Standard American Diet) to AIP in SIX” group health coaching program. Food eliminations, additions, and lifestyle modifications were done in weekly stages. Foods eliminated included all grains, legumes, nightshades, dairy, eggs, coffee, alcohol, nuts, seeds, refined/ultra-processed sugars, oils, and food additives. Micronutrient-dense food additions included foods rich in mono and polyunsaturated fatty acids, bone broth, seafood, fermented foods, and organ meats. Lifestyle modifications included the promotion of support systems, sleep hygiene, stress management, movement, and increasing time spent outdoors.
Certified health coaches and NTPs led the dietary and lifestyle intervention, educating participants with reasons for food eliminations, additions, and particular lifestyle modifications. They provided plans to help participants sustain the rigorous elimination process such as menu planning, grocery shopping, cooking foods, and recipe guides. All of the education and support was provided virtually via email and a private Facebook group accessible only to invited members. The health coaches and NTPs led daily discussions on the changes participants were implementing, provided encouragement when participants faced challenges, answered questions regarding the study process, and troubleshot with participants who experienced difficulty with the protocol.
For the purposes of the study, the health coaches and NTPs also collaborated with the lead physician in the event of any medical concerns for study participants. Through lab testing, FFQ, MSQ, and SF-36, the physician was able to recognize specific issues that individual participants were experiencing and met with health coaches to discuss methods of addressing the issues within the study framework. Where appropriate, the lead health coach and physician discussed with individual participants regarding any concerns and helped the participant address them effectively.
At the end of the intervention, questionnaires and all laboratory work, including organic acid and stool samples, were repeated. The study was conducted in full accordance with the Valley Health Research Policies and Procedures and all applicable Federal and State laws and regulations, including 45 CFR 46, 21 CFR parts 50, 54, 312, 314, and 812, as well as the Good Clinical Practice: Consolidated Guideline approved by the International Conference on Harmonisation. Participants were allowed to drop out of the study at any time.
Data collection, analysis, and outcomes
A per-protocol analysis was conducted using data from participants completing the study in its entirety (n = 16). Individuals (n = 2) who decreased thyroid medications during the study were not included in the final group analyses of thyroid hormone parameters but were included in the analyses of thyroid antibodies, including TPO antibodies and TGA. Individuals (n = 2) who were acutely ill during either the pre-intervention or the post-intervention laboratory testing were not included in the final group analyses of thyroid hormone parameters, thyroid antibodies, hs-CRP, white blood cell (WBC) count, or differential cell count analysis. Data from all 17 participants completing pre-intervention testing and 16 participants completing post-intervention testing are included in the Appendix. Specific denotations are listed in the Appendix to designate the specific data described above that was not included in the respective per protocol analyses as well as significant outlying data that was not included in the post-hoc secondary analyses.
Paired t-tests were calculated for all SF-36, MSQ, thyroid parameters, including antibodies, WBC count, differential cell counts, hs-CRP, self-reported weight, and BMI results from pre- to post-dietary intervention using Prism 8 (GraphPad Software, CA, US), resulting in a total of 27 tested parameters. It was noted during initial statistical calculations that several individual subscales of the SF-36 failed the Shapiro Wilk test for normality, and thus, all SF-36 data sets could not be assumed to be normally distributed. Wilcoxon Signed-Rank tests were thus performed for all eight subscales of the SF-36, and the respective median values were calculated and recorded alongside the respective inter-quartile range (IQR). All other data sets were assumed to be normally distributed, with statistics from the resulting paired t-tests represented as a mean (M) and standard deviation (SD). Effect sizes for normally distributed samples were also calculated using Hedge's g statistic (g) and are listed when appropriate.
In order to correct for error when performing statistical analyses for multiple hypotheses, balancing the risk of creating both Type I and Type II errors, the study authors utilized a false discovery rate control adjustment outlined by Glickman, Rao, and Schultz with a maximum false discovery rate d = 0.05 for n = 27 statistical tests. As part of this adjustment, new thresholds for statistical significance were set and are listed with their originally calculated and corresponding p-value in the Appendix [20].
The study's primary outcome was a significant change in SF-36 measures. The study's secondary outcomes consisted of changes in clinical symptom burden as measured by the MSQ, changes in thyroid parameters, including thyroid antibodies, changes in WBC and differential cell counts, and changes in hs-CRP. Measures from the organic acid and stool testing were exploratory in nature, however, pre-intervention data from these tests were utilized to inform specific dietary recommendations for individuals during week five of the intervention. These recommendations varied and were based on aspects of the organic acid test suggesting deficits in B vitamins or minerals such as magnesium, copper, riboflavin, B6, folate, or B12 as well as aspects of the stool testing suggesting overgrowth of bacterial organisms, fat malabsorption, or pancreatic insufficiency. Clinically relevant specifics of the stool and organic acid testing from individuals pre and post-intervention, as well as clinical recommendations provided midway through the 10-week intervention, are discussed as part of participant case summaries in the Appendix. Adverse effects were monitored throughout the study and recorded.
Results
Seventeen women meeting the study’s inclusion and exclusion criteria were enrolled and completed the two-week washout period. Baseline demographics, including age, height, weight, BMI, and ethnicity, are listed in Table 1. Fifteen out of 17 (87.5%) of the women were noted to be Caucasian. One participant became pregnant during the study and, as a result, discontinued participation in the study and was not included in the final analysis.
Table 1
Baseline demographics of participants included for final analysis
y (years), in (inches), lbs (pounds), BMI (body mass index), SD (standard deviation), N (sample size)
| Variable | N | Mean (SD) |
| Age, y | 16 | 35.6(5.7) |
| Height (in) | 16 | 65.3(2.4) |
| Weight (lbs) | 16 | 149.5(19.5) |
| BMI | 16 | 24.9(2.6) |
Sixteen women (n = 16) completed the SF-36 and MSQ before and after the 10-week program. There was a statistically significant improvement in HRQL as measured by all eight subscales of the SF-36 (Table 2) with the most marked improvements noted in the physical role functioning subscale with a pre-intervention median = 25, IQR 88, and post-intervention median = 100, IQR 50 (p = 0.001), the vitality subscale with a pre-intervention median = 23, IQR 19, and post-intervention median = 58, IQR 34, p < 0.0001, and the general health subscale with a pre-intervention median = 40, IQR 26, and post-intervention median = 70, IQR 35 (p < 0.0001)
Table 2
SF-36 paired t-tests results and statistics
SF-36 (36-Item Short Form Health Survey), Pre (pre-intervention), Post (post-intervention), N (sample size), IQR (inter-quartile range), P (p value), (*) denotes statistically significant p value
| SF-36 Physical Functioning | SF-36 Physical Role Functioning | SF-36 Emotional Role Functioning | SF-36 Vitality | SF-36 Mental Health | SF-36 Social Role Functioning | SF-36 Bodily Pain | SF-36 General Health | |
| N | 16 | 16 | 16 | 16 | 16 | 16 | 16 | 16 |
| Median (IQR) Pre | 80 (29) | 25 (88) | 33 (92) | 23 (19) | 54 (25) | 63 (22) | 68 (22) | 40 (26) |
| Median (IQR) Post | 95 (10) | 100 (50) | 78 (19) | 58 (34) | 78 (19) | 81 (22) | 78 (21) | 70 (35) |
| Median of Differences (IQR) | 10 (10) | 50 (75) | 41 (67) | 33 (29) | 22 (12) | 19 (37) | 23 (32) | 28 (21) |
| P | 0.0001* | 0.001* | 0.0063* | <0.0001* | <0.0001* | 0.0057* | 0.0112* | <0.0001* |
Figure 1 displays a scatter plot of SF-36 Physical Role Functioning scores pre- and post-intervention. Individual pre-intervention scores are depicted with circles and individual post-intervention scores are depicted with triangles. Error bars indicate the inter-quartile range (IQR). Solid, bolded vertical lines within the IQR indicate the median.
Figure 1
SF-36 physical role functioning scores pre- and post-intervention
SF-36 (36-Item Short Form Health Survey), PRF (physical role functioning), Pre (pre-intervention), Post (post-intervention)
Figure 2 displays a scatter plot of SF-36 physical functioning scores pre- and post-intervention. Individual pre-intervention scores are depicted with circles and individual post-intervention scores are depicted with triangles. Error bars indicate the inter-quartile range (IQR). Solid, bolded vertical lines within the IQR indicate the median.
Figure 2
SF-36 physical functioning scores pre- and post-intervention
SF-36 (36-Item Short Form Health Survey), PF (physical functioning), Pre (pre-intervention), Post (post-intervention)
Figure 3 displays a scatter plot of SF-36 vitality scores pre- and post-intervention. Individual pre-intervention scores are depicted with circles and individual post-intervention scores are depicted with triangles. Error bars indicate the inter-quartile range (IQR). Solid, bolded vertical lines within the IQR indicate the median.
Figure 3
SF-36 vitality scores pre- and post-intervention
SF-36 (36-Item Short Form Health Survey), Pre (pre-intervention), Post (post-intervention)
Figure 4 displays a scatter plot of SF-36 general health scores pre- and post-intervention. Individual pre-intervention scores are depicted with circles and individual post-intervention scores are depicted with triangles. Error bars indicate the inter-quartile range (IQR). Solid, bolded vertical lines within the IQR indicate the median.
Figure 4
SF-36 general health scores pre- and post-intervention
SF-36 (36-Item Short Form Health Survey), Gen. Health (general health), Pre (pre-intervention), Post (post-intervention)
The clinical symptom burden as determined by MSQ (Figure 5), which measures symptoms over a four-week period, decreased significantly from pre-intervention (M = 92, SD 25) to post-intervention (M = 29, SD 20), n = 16, t(15) = 9.3, p < 0.0001 with a large effect size (g = 2.81).
Figure 5
MSQ scores pre-intervention to post-intervention
MSQ (Medical Symptoms Questionnaire), Pre (pre-intervention), Post (post-intervention), SD (standard deviation), error bars indicate SD
Inflammation, as measured by hs-CRP (Figure 6), decreased significantly from pre-intervention (M = 1.63 mg/L, SD 1.72) to post-intervention (M = 1.15 mg/L, SD 1.31), n = 14, t(13) = 2.60, p = 0.0219 with a small effect size (g = 0.302). As previously noted, data from two participants who were acutely sick during either the pre- or post-intervention blood chemistry testing were not included in the final analysis for hs-CRP.
Figure 6
Serum hs-CRP from pre-intervention to post-intervention
hs-CRP (high sensitivity C-reactive protein), Pre (pre-intervention), Post (post-intervention), SD (standard deviation), error bars indicate SD
It was additionally noted when performing the statistical analysis that one participant had a significantly elevated hs-CRP both pre-and post-intervention when compared to the pre- and post-intervention group means, however, she was not acutely sick during either the pre- or post-intervention blood chemistry testing. While the participant’s hs-CRP was noted to decrease from pre- to post-intervention, her data still remained a significant outlier from the group mean as seen in the previous scatter plot (Figure 6). A post-hoc secondary analysis was conducted removing the statistical outlier, resulting in a sample size of n = 13, a pre-intervention mean, M = 1.22 mg/L, SD 0.81, and post-intervention mean, M = 0.85 mg/L, SD 0.72, t(12) = 2.34, p = 0.037 with a moderate effect size (g = 0.473). The pre- and post-intervention hs-CRP data tables in the Appendix denote the specific data from the two acutely sick individuals described above that was not included in the final analysis as well as the data from the outlier that was not included in the post-hoc secondary analysis.
Pre- and post-statistics for all thyroid markers, including antibodies, are listed in Table 3. Individuals who decreased medication use following initial laboratory testing or during the course of the study (n = 2), as described previously in the methods, were not included in the final analysis. Additionally, data from the two participants (n = 2) who were acutely sick during the pre- or post-intervention thyroid testing were not included in the final analysis for TSH, free T4 and T3, total T3 and T4, and reverse T3. Data regarding antibody levels, however, for these two participants were included in the final data analysis. This resulted in a total of 12 participants analyzed for thyroid markers and 14 analyzed for thyroid antibodies. All data for the 17 participants completing pre-intervention thyroid testing as well as the 16 participants completing post-intervention thyroid testing is listed in the Appendix with specific denotations for the individual data described above that was not included in the final data analysis.
Table 3
Thyroid hormone and antibody values pre- and post-intervention with paired t-test statistics
TPO (thyroid peroxidase antibodies), TGA (anti-thyroglobulin antibodies), pre (pre-intervention), post (post-intervention), N (sample size), SD (standard deviation), t (t-test statistic), P (p-value), g (Hedges' g)
| TSH (μIU/mL) | Total T3 (ng/dL) | Free T3 (pg/mL) | Reverse T3 (ng/dL) | Total T4 (μg/dL) | Free T4 (ng/dL) | TPO (IU/mL) | TGA (IU/mL) | |
| N | 12 | 12 | 12 | 12 | 12 | 12 | 14 | 14 |
| Mean (SD) pre | 2.02(1.46) | 97.3(18.0) | 2.4(0.6) | 17.4(4.3) | 7.0(1.1) | 1.3(0.4) | 225(178) | 110(261) |
| Mean (SD) post | 1.98(1.44) | 89.0(9.0) | 2.4(0.5) | 19.1 (5.3) | 7.1(1.4) | 1.4(0.4) | 219(186) | 124(293) |
| t | 0.075 | 1.668 | 0.1515 | 1.9717 | 0.5932 | 0.841 | 0.7703 | 1.4292 |
| P | 0.942 | 0.124 | 0.882 | 0.0743 | 0.565 | 0.418 | 0.455 | 0.176 |
| g | 0.029 | 0.584 | 0.029 | 0.355 | 0.124 | 0.099 | 0.035 | 0.0532 |
No clinically nor statistically significant changes were seen in TSH, total T3 or T4, and free T3 or T4. Additionally, no clinically nor statistically significant changes were noted for either TPO antibodies or TGA.
White blood cell (WBC) and differential cell counts pre- and post-intervention are listed in Table 4. It was noted that there was a decrease in mean WBC count from a pre-intervention mean of 5.6 x 103 / μL (SD 1.4) to a post-intervention mean of 5.1 x 103/μL (SD 1.4) that did not reach statistical significance, p = 0.1396. As previously noted, two out of the 16 individuals completing the pre- and post-intervention blood chemistry were noted to be acutely sick during either the pre- or post-intervention laboratory testing period and could not be included in the final analysis.
Table 4
WBC and differential cell counts pre- and post-intervention with paired t-test statistics
WBC (white blood cell), Pre (pre-intervention), Post (post-intervention), N (sample size), SD (standard deviation), t (t-test statistic), P (p-value), g (Hedges' g)
| WBC (103/μL) | Neutrophils (%) | Lymphocytes (%) | Monocytes (%) | Eosinophils (%) | |
| N | 14 | 14 | 14 | 14 | 14 |
| Mean (SD) Pre | 5.6(1.4) | 57.9(6.6) | 30.8(5.3) | 8.4(2.3) | 2.4(1.6) |
| Mean (SD) Post | 5.1(1.4) | 55.9(8.2) | 34.0(7.3) | 7.4(1.3) | 2.1(1.4) |
| P | 0.1396 | 0.183 | 0.0286 | 0.0684 | 0.385 |
| g | 0.311 | 0.268 | 0.502 | 0.535 | 0.199 |
Figure 7 displays a box plot depicting WBC counts both pre- and post-intervention as well as the mean WBC count with SD. It was noted when performing the paired t-test statistics and creating the box plot that one individual was a significant outlier when compared to the group mean difference in WBC count with an increase in WBC from 6.3 x 103/μL pre-intervention to 8.4 x 103/μL post-intervention. The box plot also depicts two participants who began the intervention with low or borderline low WBC counts (normal > 3.3 x 103/μL) and had increases in WBC count at post-intervention trending toward the group post-intervention mean.
Figure 7
WBC count pre- and post-intervention
WBC (white blood cell), Pre (pre-intervention), Post (post-intervention), error bars indicate SD
Statistically significant changes were seen in weight (M = 143.4 lbs, SD 16.7, p = 0.002) and BMI (M = 23.9, SD 2.2, p = 0.002) from baseline to post-intervention (Table 5).
Table 5
Weight and BMI of all participants pre- and post-intervention, (*) denotes statistically significant P value (p < 0.05)
BMI (body mass index), lbs (pounds), N (sample size), SD (standard deviation), Pre (pre-intervention), Post (post-intervention), P (p-value)
| Weight (lbs) | BMI | |
| N | 16 | 16 |
| Mean (SD) Pre | 149.5(19.5) | 24.9(2.6) |
| Mean (SD) Post | 143.4(16.7) | 23.9(2.2) |
| P | 0.002* | 0.002* |
These results remained significant when a subgroup analysis was performed on participants with a baseline BMI ≥ 25 (M = 27.1, SD 1.0) to post intervention (M = 25.8, SD 1.0, p = 0.011) (Table 6).
Table 6
Weight and BMI of overweight subjects (24.9 < BMI < 29.9) pre- and post-intervention, (*) denotes statistically significant P value (p < 0.05)
BMI (body mass index), lbs (pounds), N (sample size), SD (standard deviation), Pre (pre-intervention), Post (post-intervention), P (p-value)
| Weight (lbs) | BMI | |
| N | 8 | 8 |
| Mean (SD) Pre | 163.0(19.1) | 27.1(1.0) |
| Mean (SD) Post | 155.1(15.3) | 25.8(1.0) |
| P | 0.011* | 0.011* |
It should be noted that the review of the individual FFQs was used to determine qualitative and compositional changes in dietary habits with regards to eliminated foods as part of AIP and was not used to determine portion sizes or total caloric intake.
Six out of the 13 women beginning the study on thyroid replacement medication decreased their dose of hormone replacement medication after the 10-week intervention. All three individuals who decreased the dosages of their medication following the pre-intervention testing made subsequent decreases in their medication dosage in addition to three individuals who decreased their medication dosages following post-intervention laboratory testing. All three of the women who began the study without the use of hormone replacement medication continued without the use of replacement medication as of the final post-intervention study visit.
There were no moderate to severe adverse effects noted during the duration of the study. Some study participants reported mental challenges during the initial phases of the dietary eliminations, however, this is was offset very quickly by decreases in overall symptom burden.
Discussion
This single-arm pilot study adds to the current evidence that AIP, a modification of the Paleolithic diet involving the elimination and promotion of certain foods, may help alleviate symptoms and improve quality of life in participants with an autoimmune disease. We demonstrated preliminary efficacy in participants with HT, via statistically and clinically significant improvements in SF-36 and MSQ scores, as well as statistically and clinically significant decreases in hs-CRP, weight, and BMI despite no statistically significant changes in thyroid laboratory markers or thyroid antibodies. Reviews of FFQs from participants during the 10-week program revealed 95%-100% adherence to the strict elimination criteria. The strict dietary adherence is most likely a result of the intensive health coaching and community-based structure providing both education and a source of communal accountability. Preliminary study questionnaires revealed a majority of the participants reporting familiarity with AIP. Some participants even reported previous attempts at the AIP dietary protocol for fewer than five days, given the lack of education about the dietary approach, support services, and communal accountability as well as the overall challenge in preparing 100% AIP-compliant meals. The role of the physician, health coaches, and NTPs, as well as the participants’ communal group environment, cannot be understated and appears to be the primary mediating elements behind the high rate of adherence.
These results additionally suggest that the AIP diet and concomitant lifestyle modification, as implemented by a multi-disciplinary team, can be safely used as adjunctive treatments for people with HT who are already utilizing hormone replacement therapy. There were no reported serious adverse effects, with many participants actually reporting noticeable positive changes within the first four weeks of the elimination diet. While there were no observed changes in mean thyroid laboratory markers and antibodies, six out of 13 women (46.1%) who were taking thyroid replacement medication at the beginning of the study actually decreased their dosage of hormone replacement medication by the end of the 10-week study period. All three women who were asked to decrease or alter their medication dosing at the beginning of the study due to pre-intervention laboratory findings of low TSH or abnormal free hormone levels actually found they needed to decrease their medications even further following the 10-week program. Three women who began the study without the utilization of hormone replacement medication were able to continue without hormone replacement medication. One individual who enrolled with subclinical hypothyroidism and elevated thyroid antibodies diagnostic of HT had a significantly higher post-intervention TSH, yet nearly identical free and total hormone levels as well as lower TPO antibodies at post-intervention. It is difficult to predict the continued disease course of this specific individual outside of the study structure, however, it is likely that she would require both hormonal replacement therapy with concomitant dietary and lifestyle support to manage any further progression of autoimmune thyroiditis.
Despite the lack of a significant clinical change in mean thyroid markers, including antibodies, the statistically and clinically significant decreases observed for hs-CRP point to modulation of the overall immune and inflammatory response underlying autoimmune thyroiditis. Additionally, a secondary post-hoc analysis of changes in WBC count from pre- to post-intervention (n = 13) that removed one statistical outlier resulted in a statistically significant change in mean WBC count from 5.5 x 103/μL (SD 1.4) to 4.9 x 103/μL (SD 1.1), p = 0.0205. There was also a noted increase in mean lymphocyte count from pre- to post-intervention (p = 0.0286) that could not be assumed to be statistically significant after correcting for multiple hypothesis testing using the false discovery rate correction. It is clear from these statistical examinations that there is some underlying modulation of the immune system that is not as statistically or clinically robust as the changes in HRQL and clinical symptom burden but, nonetheless, should be acknowledged and explored in further study of the AIP dietary intervention. It is also interesting to note that the pre-intervention mean hs-CRP in the post-hoc secondary analysis that included n = 13 subjects was classified as intermediate risk while the post-intervention mean actually dropped below 1.0 mg/L into low-risk categorization.
The authors speculate that it is possible that one would observe an eventual decrease in thyroid antibodies and a decreased need for supplemental medication as well as more robust changes in immune and inflammatory markers in participants adhering to the AIP dietary principles for additional periods of six to 12 months.
In speculating as to the mechanisms behind the observed positive changes in quality of life, symptom burden, as well as hs-CRP, we suggest a further examination of the original criteria set forth for a feasible and efficacious dietary and lifestyle intervention. Self-reported FFQs and dietary journals provided throughout the duration of the study indicate the inclusion of foods with greater nutrient density by all participants and the exclusion of less nutrient-dense foods. Qualitative post-intervention surveys additionally appear to indicate that the study participants received a positive benefit from the gradual nature of the dietary eliminations, the consistent support from the multi-disciplinary team, and the ability to interact with other participants making the same dietary and lifestyle changes. There was a statistically and clinically significant change in weight and BMI from pre- to post-intervention within both the cohort as well as a smaller overweight subpopulation. Despite the dietary intervention lacking a focus on caloric quantification, the restriction of specific macronutrients, such as carbohydrates or fats, or an emphasis on the promotion of weight loss, individuals indicated self-reported weight loss from pre- to post-intervention that likely contributed to improvements in HRQL and symptom burden.
While the study program is inherently confounded due to its multi-faceted design, including social support, lifestyle education, medical supervision, and dietary guidance, the profound improvements observed in the quality of life and symptom burden seem to indicate a synergistic and compounding benefit from the inclusion of multiple therapeutic elements.
There are currently no published studies assessing the utilization of a comprehensive dietary and lifestyle intervention in participants with HT, making it difficult to provide an analysis of comparative or expected treatment effects. Winther et al. assessed the role of thyroxine treatment over a six-month period to improve quality of life in a population of 78 individuals newly diagnosed with HT and either subclinical hypothyroidism (4 µIU/mL < TSH < 10 µIU/mL), n = 66, or overt hypothyroidism (TSH > 10 µIU/mL ), n = 12 [3]. The authors noted that despite optimal medical management over the six-month period, only the SF-36 subscales of vitality, physical role functioning, and mental health showed statistically significant changes [3]. When examining the data from the 58 participants completing the six-month study, it should be noted that the increases in these three domains ranged from only 3%-8%, correlating to a minimal effect size as well as a potentially insignificant change in clinical status [3].
While we cannot compare SF-36 statistics from the AIP intervention directly with those from Winther et al., as we could not assume all SF 36 subscale data sets from the AIP intervention to be normally distributed and thus could not accurately calculate respective means and SDs, it is worth examining some of the more notable pre- to post-intervention changes for specific SF-36 subscales between the AIP intervention and the study group from Winther et al.
In examining the relative magnitude of changes for the SF-36 subscales: physical role functioning, vitality, mental health, and general health from the current intervention, the study authors observed a remarkable increase in physical role functioning scores from a median of 25, IQR 88, pre-intervention to a post-intervention median of 100, IQR 50, corresponding to a median difference of 50, IQR 75. Similar large-magnitude changes were noted when examining median values pre- and post-intervention for the vitality, mental health, and general health subscales. The pre-intervention median vitality subscale score was noted to be 23, IQR 19, however, post-intervention, the median vitality subscale score increased to 58, IQR 34, with a median difference of 33, IQR 29. The pre-intervention median mental health subscale score was noted to be 54, IQR 25, however, post-intervention, the median mental health subscale score increased to 78, IQR 19, with a median difference of 22, IQR 12. The pre-intervention median general health subscale score was noted to be 40, IQR 26, however, post-intervention, the median general health subscale score increased to 70, IQR 35, with a median difference of 28, IQR 21.
When comparing the magnitude of change noted for these three SF-36 subscales between the study from Winther et al. and the AIP intervention, it is important, first, to note the small and underpowered sample size of the AIP study. Additionally, the vitality subscale scores in the AIP trial at baseline were lower when compared to the baseline scores of participants in the Winther et al. trial, with both study populations becoming clinically equivalent post-intervention.
Mental health scores in this trial at baseline were clinically similar to those of Winther et al. (AIP pre-intervention median = 54, IQR 25; Winter et al. pre-intervention mean = 47 (SD 9), however, at post-intervention, there was a marked difference between these two study groups with the median mental health score post-intervention from the AIP study being equal to 78 (IQR 19) while the mean mental health score post-intervention from the Winther et al. trial was 50 (SD 10) [3]. When comparing the other SF-36 domains, greater improvements were also seen in the physical functioning, bodily pain, emotional role functioning, and social role functioning scales for participants in the AIP trial as compared to those in the Winther et al. trial [3].
In examining the bodily pain SF-36 subscale in the AIP trial, there was a notable increase from a pre-intervention median of 68 (IQR 22) to 78 (21) at post-intervention (p = 0.0112) as compared to Winther et al.: 52 (SD 12) pre-intervention to 55 (SD 10), p > 0.05 following six months of levothyroxine therapy [3]. In reviewing the specific subscales of the MSQ, it appeared that the improvements seen in bodily pain as measured by the SF-36 were primarily related to improvements in joint pain, muscle aches, and headaches as indicated more specifically by scores from the MSQ. Given the current concerns surrounding opioid misuse/overuse in those suffering from chronic pain conditions, including individuals with HT, multi-dimensional, non-pharmacologic interventions, such as the AIP dietary and lifestyle intervention utilized in this trial, may provide clinicians with novel, efficacious, and low-risk treatments for chronic pain.
Improvements in quality of life and symptom burden may be of critical benefit for both patients and clinicians, as it may help increase trust in providers as well as adherence to continued medical and lifestyle therapy. Despite prior work indicating that quality of life could be negatively impacted by restrictive diets, this study suggests that quality of life was not negatively impacted but markedly enhanced [10].
The reason for our conflicting findings regarding changes in quality of life, as previously suggested, may be due to the AIP study’s multi-dimensional treatment design involving frequent monitoring and interactions between participants with the team of health coaches and NTPs via a private Facebook group. Research on cancer survivors has shown preliminary evidence linking increased fruit and vegetable intake to increased social support as well as feelings of hope, possibly indicating a mechanism by which social support alone can improve one’s overall food choices [21].
Additional research indicates a strong association between a person’s quantity and quality of social interactions and their perceived health and quality of life [22]. It is unlikely, however, that given the profound improvements in the physical role functioning as well as vitality and general health subscales of the SF-36 that social interaction alone, whether between study participants themselves or between study participants and the multi-disciplinary team could account for all of the observed improvements in quality of life as well as overall symptom burden.
Limitations to the study include its small sample size, the lack of a control group, the lack of blinding, the possibility for selection bias of participants enrolling in the study, as well as response bias from participants regarding their weights. Additional limitations include the use of a medical symptoms questionnaire that has yet to be validated in large populations as well as the potentially transient nature of the participant’s symptoms being documented by the questionnaire. Further limitations to this study include the lack of data collection on physical activity, sleep, social support, stress management, or the effect that eliminated foods would have had if they were to be reintroduced systematically.
Conclusions
Our pilot study suggests that an online, community-based AIP diet and lifestyle program facilitated by a multi-disciplinary team can significantly improve HRQL and symptom burden in middle-aged female subjects with HT. While there were no statistically significant changes noted in thyroid function or thyroid antibodies, the study’s findings suggest that AIP may decrease systemic inflammation and modulate the immune system, as evidenced by the decreases in average hs-CRP. Dietary and lifestyle changes may be a significant life stressor, but the use of health coaches and NTPs, in addition to nutritionally trained physicians, may offset this and provide an increase in quality of life. Larger randomized controlled trials are necessary to validate these findings and examine long-term follow-up, adherence, and any adverse events during the elimination and/or maintenance phase of AIP. Given the low-risk nature of the AIP dietary and lifestyle intervention as well as the improvements seen in HRQL and the participants’ symptom burden, further study in larger populations of individuals with HT implementing AIP as part of a multi-disciplinary diet and lifestyle program is warranted.
Copyright © 2019, Abbott et al. This work is published under http://creativecommons.org/licenses/by/3.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.