Efficacy of an adjuvant Lactobacillus rhamnosus formula in improving skin lesions as assessed by PASI in patients with plaque psoriasis from a university-affiliated, tertiary-referral hospital in São Paulo (Brazil): a parallel, double-blind, randomized clinical trial

Psoriasis is an inflammatory disease of the skin, characterized by erythematous plaques. It is rather common, affecting 2–4% of the population in western countries. Psoriasis’ etiology encompasses both genetic and environmental factors. Evidence suggests that the latter reflect the importance of changes in the microbiome for developing the disease. Thus, it is hypothesized that gut microbiome manipulation may arise as a way of treating psoriasis. However, few trials assessed the use of probiotics in psoriasis, although promising results were detected in small studies. Our objective was to assess the efficacy of adjuvant probiotics (Lactobacillus rhamnosus) in treating plaque psoriasis patients. This was a randomized, parallel, placebo-controlled, double-blind trial with two arms: experimental (n = 50) and control (n = 53). Inclusion of subjects and data gathering lasted from November 2020 to August 2021. Subjects were consecutive plaque psoriasis patients under regular follow-up in the Dermatology unit of a university-affiliated, tertiary-referral hospital in São Paulo (Brazil). Eligibility criteria included being over 18 years old, having plaque psoriasis and not having other skin diseases, neoplasms nor systemic inflammatory diseases. Subjects received standard-of-care plus probiotics (Lactobacillus rhamnosus formula). Controls received standard-of-care plus placebo. Primary outcome was skin lesion improvement as assessed by psoriasis area of severity index (PASI) at six months. Secondary outcome was quality-of-life as assessed by dermatology life quality index (DLQI) at six months. Regarding within-group analyses, changes in both PASI and DLQI were non-significant for the experimental group (mean PASI decreased by 1.58, p = 0.105, and mean DLQI increased by 0.05, p = 0.873) and significant for controls (mean PASI decreased by 1.90, p = 0.019, and mean DLQI decreased by 3.33, p = 0.031). Between-group analyses returned non-significant results (p = 0.620). Our findings do not support the hypothesis that gut microbiome modulation via ingestion of Lactobacillus rhamnosus produces clinical improvement in psoriasis patients. Further research is encouraged. Trial registration: Retrospectively registered at the Brazilian Clinical Trials Registry (RBR-8js7t83) on 08/02/2022.


Introduction
Psoriasis is a chronic, immune-mediated, inflammatory disease of the skin and joints, characterized by well-circumscribed, erythematous plaques [1]. It affects roughly 2-4% of western countries' population [2], with a prevalence of around 2% in the United States [1]. Its most common phenotype is plaque psoriasis (90% of cases) [3]. Psoriasis is associated with an increased risk of mortality and comorbidities (e.g., hypertension, obesity and depression), and its inflammatory process is a risk factor for developing diseases related to cardiovascular and metabolic dysregulation [4].
Although psoriasis' pathogenesis is not fully understood, it is multifactorial and encompasses genetic and environmental factors. It is hypothesized that the latter reflects the importance of changes in the microbiome (set of microbes that live in association with the human body) for developing the disease [5,6], which is reinforced by two findings: the gut microbiome of psoriasis patients differs from that of healthy patients (e.g., the Firmicutes/ Bacteroidetes ratio and Actinobacteria levels are lower in psoriasis) [7][8][9], and factors that alter gut microorganisms' activity were effective in modulating the chronic inflammation of certain skin diseases [10].
Furthermore, evidence increasingly suggests that modulating gut microbe communities' composition via probiotics can induce immune responses that affect other body parts-like the skin [11]. Thus, it is hypothesized that gut microbiome manipulation may be a way of treating diseases [12]. However, this approach lacks in-depth research. Few trials assessed the use of probiotics in psoriasis, although promising results were detected in small studies [13,14] and in the context of atopic dermatitis [15,16].

Objectives
To assess the efficacy of adjuvant probiotics (Lactobacillus rhamnosus) in treating plaque psoriasis patients.

Methods
This was a randomized, parallel, placebo-controlled, double-blind, 6-month long trial with two arms (experimental and control), retrospectively registered at the Brazilian Clinical Trials Registry (RBR-8js7t83) on 08/02/2022. The research protocol was approved by our Institutional Review Board on 09/24/2019 (16435719.3.0000.5479). CONSORT's reporting guidelines [17] were followed. All methods remained unchanged throughout the trial. All data generated or analyzed during this study are included in this article and its supplements (Online Resource 3).
Subjects were consecutive patients under regular followup in the Dermatology unit of a university-affiliated, tertiaryreferral hospital in São Paulo (Brazil). Eligibility criteria included being over 18 years old, having plaque psoriasis, not being pregnant and not having other skin diseases, neoplasms nor systemic inflammatory diseases (such as Crohn's disease and inflammatory bowel disease). To preserve realworld circumstances and increase results' generalizability, type of ongoing psoriasis treatment was not used as exclusion criterion (also, subgroup analyses were performed to stratify by type of treatment and avoid confounding). In accordance with the Declaration of Helsinki, all participants provided written informed consent for both participation and data publishing before inclusion. Consent is stored on file and available upon request.
Participants who consented and met the eligibility criteria were allocated to either group following a 1:1 ratio, via permuted-block randomization (blocks of different sizes) performed on a specialized website. To preserve allocation concealment, opaque sequentially-numbered envelopes were used and block sizes were not disclosed to clinicians. These processes were performed by an investigator unrelated to data gathering and outcome assessment.
Subjects, treatment deliverers and outcome assessors remained blinded throughout the trial. To preserve blinding, probiotics and placebo vials looked exactly the same. Unblinding was permissible under emergencies, which were not expected due to the intervention's good safety profile. No cases of unblinding occurred.
To improve adherence to protocol, participants were routinely contacted via phone by principal investigators to elucidate questions and encourage compliance.
Inclusion of subjects and data gathering lasted from November 2020 to August 2021. Upon inclusion, participants were submitted to a thorough assessment of health data. The "smoking" variable followed the modified National Survey on Drug Use and Health (NSDUH-M) definition [18]. Subjects were given enough vials of probiotics or placebo for until their next appointment (scheduled for 2-3 months later, following the local protocol) and instructed on how to properly store and consume them. Upon returning, patients were reassessed and given more vials. This process was repeated until each subject had completed six months in the study.
Subjects were given standard-of-care plus probiotics (experimental group) or standard-of-care plus placebo (control group) for six months. Lactobacillus rhamnosus was chosen due to its good biosafety profile, extensive use in microbiome research and suggested efficacy in ameliorating inflammation in both gut and skin diseases [19][20][21]. Vials were composed of a 5% whey formula with or without probiotics. The formula was made by adding 1000 ml of deionized water to 50 g of whey, agitating until complete dissolution, distributing in vials (95 ml each), capping vials half-screwed, heating in autoclave at 111 °C for 15 min, waiting for cooldown, finishing capping and controlling for sterility. Probiotics were made by growing Lactobacillus rhamnosus ATCC 7469 strains in blood agar and chocolate agar, incubating at 35 °C ± 2 °C under capnophilic atmosphere for 24 h, inoculating strains in MRS agar at 35 °C ± 2 °C for 24 h and controlling for sterility in blood agar and MacConkey agar. For vials with probiotics, 1.9 µl of Lactobacillus rhamnosus were added to the 5% whey formula (resulting in 6 × 10 5 bacteria per milliliter) and incubated at 35 °C ± 2 °C for 48 h. All vials were labeled for batch, expiration date and conservation instructions. Vials were produced and donated by an external laboratory (Probac do Brasil Produtos Bacteriologicos LTDA) and stored in a refrigerator in our Dermatology unit (2-8 °C). When properly stored, vials' expiration date was six months after production. At room temperature, expiration would happen in four hours. No reports of spoiled vials occurred. Patients were instructed to store vials in common refrigerators and drink 5 ml of probiotics/placebo daily (tip-less syringes with volume markings were provided).
Primary outcome was skin lesion improvement as assessed by Psoriasis Area of Severity Index (PASI) at six months. PASI is a quantitative score for measuring the psoriatic lesions' severity, based on area coverage and plaque appearance. It has been extensively validated [22], including in Brazil [23]. Secondary outcome was quality-of-life as assessed by dermatology life quality index (DLQI) at six months, which served a more exploratory purpose. DLQI is a self-administered and validated questionnaire, designed to measure the quality-of-life of adult patients with skin diseases [24,25]. Both scores are widely applied in clinical settings and cited in the literature, presenting high levels of reliability, applicability and reproducibility [26,27]. Patients underwent PASI and DLQI assessments by blinded evaluators at each appointment.
Subjects' data were registered by investigators in paper and transcribed to an encrypted Electronic Data Capture software (REDCap ® ), maintained by our center's data managers.
For the sample size calculation, we considered: alpha = 0.05, power = 0.80 and allocation rate = 1:1. Due to scarcity of previous studies, values for expected effect difference and standard deviation (SD) were unknown. We adopted the values provided by the trial that shared most similarities with ours [28], which were: expected PASI difference between groups of 30% at six months, SD of 5.11 in the experimental arm and SD of 4.24 in the control arm. Thus, to detect a moderate effect (Cohen's d = 0.638), the sample would need to include 80 subjects. Considering a conservative drop-out rate of 25%, the final sample size was 100 subjects.
Variables were assessed for normality via histogram and Shapiro-Wilk test. Categorical variables were analyzed with Fisher's exact test. Independent continuous variables were analyzed with Student's unpaired t-test (parametric) or Mann-Whitney test (non-parametric). Exploratory linear and logistic regressions were performed to evaluate how variables affected outcomes. Limit for significance was set at p = 0.05 (two-tailed). Analyzes were conducted on Stata/ BE 17.0 (StataCorp, College Station, TX, USA). No interim analyses were conducted.
An intention-to-treat (ITT) approach was adopted to preserve randomization and avoid biases caused by eventual non-random drop-out patterns. Missing data was adjusted for baseline characteristics via multiple imputation. A sensitivity analysis (with the baseline-observation-carried-forward method, a conservative approach) was performed to assess results' robustness. Complementary, a per-protocol (PP) analysis was performed (considering a minimum adherence to intervention of 83%, which translates to five months) to further assess data robustness.

Results
A flowchart with the number of patients at different stages of the study, plus reasons for drop-outs, is presented in Fig. 1. Out of 122 screened patients, 103 were included. Adverse effects occurred in one subject from the experimental arm and four controls. Subjects' baseline characteristics are presented in Table 1. Excepting mean baseline DLQI, groups were similar in regards to all variables.
Firstly, an ITT approach was adopted. Since PASI and DLQI values exhibited non-normal distributions, non-parametric tools were used: Wilcoxon signed-rank tests (withingroup analyses) and Mann-Whitney tests (between-group analyses).
Changes were non-significant within the experimental group and significant within the control group (Table 2): mean PASI decreased by 1.58 in the intervention arm (p = 0.105) and by 1.90 in the control arm (p = 0.019), while mean DLQI increased by 0.05 in the intervention arm (p = 0.873) and decreased by 3.33 in the control arm (p = 0.031).
Between-group analyses returned non-significant results for post-intervention changes in both PASI (p = 0.620) and DLQI (p = 0.107) (Fig. 2). Stratifying subjects according to type of ongoing psoriasis treatment still returned nonsignificant results. Regarding categorical variables, controls 1 3 achieved higher rates of PASI 50, 75 and 90 than the intervention group, but these results were non-significant (Online Resource 1). Results remained non-significant after a sensitivity analysis (baseline-observation-carried-forward) and in the PP analysis, demonstrating data robustness.
Linear regressions (based on ITT) were adjusted for the most influential covariates, revealing a negative association between change in PASI and baseline PASI (p = 0.000) and a positive association between change in PASI and phototherapy (p = 0.026), with adjusted R 2 = 0.78. Remaining results were non-significant (Table 3).
Logistic regressions (based on ITT) showed significant associations between: (I) PASI 50 and baseline PASI; (II) PASI 75 and age, baseline PASI, hypertension and diabetes; and (III) PASI 90 and baseline PASI. Although statistical significance was achieved in these comparisons, some 95% confidence intervals were considerably large (Online Resource 2).

Discussion
Randomized clinical trials evaluating probiotics' effects in psoriasis patients are scarce, with only one [28] coming to our knowledge. In it, subjects were administered a mixture of Bifidobacterium longum, Bifidobacterium lactis and Lactobacillus rhamnosus and compared to controls on placebo. After 12 weeks, 66.7% of subjects achieved PASI 75, against 41.9% of controls (p = 0.032). These results greatly differ from ours, in which no significant difference in PASI 75 between groups was detected. Furthermore, our findings were unexpected: subjects from the intervention arm did not improve at all (in fact, their mean DLQI slightly deteriorated, which may have been due to chance alone), while controls significantly improved in both PASI and DLQI (although this improvement was non-significant in the between-group analysis). Controls' improvement in DLQI may have derived from their significantly higher values of baseline DLQI, the only variable that was not evenly distributed between groups. Such higher DLQI, that is hypothesized to have been due to chance alone, likely reflects a strong burden on quality-of-life, which may have rendered controls more prone to improving due to the placebo effect.
Regarding regressions' results, the negative association between change in PASI (variable in which negative values indicate improvement) and baseline PASI was expected: higher PASI indicates more severe, inflamed disease forms, which are likely to respond better to the anti-inflammatory effect that gut microbiome modulation is expected to produce. The same applies for the associations between baseline PASI and PASI 50, 75 and 90. Contrarily, the positive association between change in PASI and phototherapy lacks biological plausibility and was presumably due to chance alone. The same applies for the relationship between PASI 75, age and hypertension. Finally, the association between PASI 75 and diabetes (condition characterized by systemic inflammation) was expected. This is the second randomized clinical trial to assess probiotics' effects on plaque psoriasis patients. Due to scarcity of previous studies, some values used for our sample size calculation have little literature support, which was addressed via adoption of conservative values. This was a single-center, hospital-based study, rather than a community-based one, rendering it unclear to what extent our findings can be generalized. Also, biases related to our methods of patient/control inclusion are pertinent. Nevertheless,  there are no restrictions for patients to be admitted to our hospital, and all consecutive cases admitted were included. PASI is evaluator-dependent and DLQI is patient-dependent, both prone to individual subjectivity. Still, their use in this study was driven by their wide acceptance in Dermatology research and practice. Plus, patients and evaluators were blinded. Groups' baseline characteristics were similar in regards to all variables, excepting mean baseline DLQI. This was taken into account when interpreting results (refer to the first paragraph of this section) and was not considered a relevant issue, since DLQI was the secondary outcome and served more an exploratory purpose. In order to simulate real-world circumstances, produce a heterogeneous sample that was more representative of the target population and increase results' generalizability, type of ongoing psoriasis treatment was not used as exclusion criterion. However, that risks distorting our perception of the probiotics' effect in two ways: (a) ongoing treatment may interact with probiotics and modify their effect, or (b) patients may improve due to the ongoing treatment rather than due to probiotics. This was addressed by having subjects under different types of ongoing treatment in both groups (evenly distributed, via proper randomization) and by performing subgroup analyses that stratified subjects according to them (no significant associations were detected). For ethical reasons, subjects were not denied regular psoriasis treatment: they regularly attended our Dermatology unit and had their medications adjusted. Thus, some degree of improvement in both PASI and DLQI is expected, independently of probiotic usage, which could distort our perception of the probiotics' effect. However, such bias was at least partially mitigated by comparing subjects against controls on placebo.
Our findings do not support the hypothesis that gut microbiome modulation via ingestion of Lactobacillus rhamnosus produces clinical improvement in plaque psoriasis patients. It is unclear to what extent such results have derived from the probiotics' effect itself, our sample's characteristics or other causes. However, although previous trials on this topic were scarce, they returned promising results. Further research is needed. Future trials may benefit from enrolling larger sample sizes, using some types of ongoing psoriasis treatment as exclusion criteria, recruiting different populations and/or administering probiotics based on different bacterial strains.

Conclusions
Psoriasis can strongly impair patients' quality-of-life, not to mention the burden it represents for the Healthcare system. Therefore, we must explore therapeutic alternatives. Microbiome manipulation can be an option that is effective, inexpensive and easy to implement. Our findings do not support the hypothesis that gut microbiome modulation via ingestion of Lactobacillus rhamnosus produces clinical improvement in plaque psoriasis patients. Still, these results may signal a potentially modest effect of Lactobacillus rhamnosus supplementation on psoriasis patients or an effect that is more pronounced in specific subgroups. Furthermore, these findings do not exclude the possibility that probiotics based on other bacterial strains may produce clinically relevant effects in psoriasis patients. Further research is encouraged.

Supplementary Information
The online version contains supplementary material available at https:// doi. org/ 10. 1007/ s00403-023-02553-1. Availability of data and materials All data generated or analyzed during this study are included in this article and its supplementary files.

Declarations
Conflict of interest All authors state that there are no conflicts of interest to disclose.

Ethics approval
The research protocol was approved by our Institutional Review Board on 09/24/2019 (under the registration number 16435719.3.0000.5479).
Informed consent In accordance with the Declaration of Helsinki, all participants provided written informed consent before inclusion, for both participation and data publication. Patient consent is stored on file and readily available upon request.