**4. Discussion**

The effect of OLZ on the microbiota has been investigated in experimental studies. Davey et al. [29,30] found decreased gu<sup>t</sup> microbiota diversity, increased abundance of phyla Firmicutes, and reduced Actinobacteria, Proteobacteria, and Bacteroidetes in the course of OLZ treatment in female rats. Similarly, Morgan et al. [31] revealed decreased alpha diversity, lower abundance of class

Bacteroidia, and increased abundances of Erysipelotrichia, Actinobacteria, and Gammaproteobacteria in female mice treated with OLZ. However, Kao et al. [44] demonstrated no significant effects of OLZ on gu<sup>t</sup> microbiota in female rats. To the best of our knowledge, this study is the first to analyze fecal microbiota compositions in patients hospitalized due to acute relapse of SZ. We did not find the impact of six-week OLZ treatment on bacterial diversity, abundance, and predicted metabolic function, and patients with SZ had individualized and stable gu<sup>t</sup> microbiota in the course of six-week OLZ treatment in terms of both composition and function. Because of the inconsistent findings above, further studies are needed to clarify the effect of OLZ on gu<sup>t</sup> microbiota.

Although gu<sup>t</sup> microbiota could be compositionally and functionally clustered into similar groups, the classification could not be used to predict the responses to OLZ treatment or the occurrence of weight gain (observed only in women) during OLZ treatment. As mentioned above, OLZ causes weight gain in female rats [29,44] and mice [31]. This metabolic effect is not observed during antibiotic therapy [29] and gnotobiosis (germ-free mouse model) and is enhanced during the administration of the high-fat diet regimen that is responsible for alterations of microbiota similar to those observed in metabolic syndromes [31]. In addition, Davey et al. [30] demonstrated metabolic disturbances, inflammation, and microbiota alterations in female mice treated with OLZ and found only slight alterations in male mice treated with OLZ, and metabolic effects of OLZ were linked to gu<sup>t</sup> microbiota alterations. Notably, antibiotics reversed these effects due to reduced gu<sup>t</sup> microbiota. Therefore, changed gu<sup>t</sup> microbiota plays a pivotal role in weight gain. The lack of association between fecal microbiota compositions and weight gain in this study may be due to the low number of participants and the short period of OLZ administration. In addition, other factors might also be responsible for the increase in body mass index associated with the OLZ administration [20,26–28].

In the present study two bacterial enterotypes (clusters) were found, Type 1, with a predominance of *Prevotella*, and Type 2 with a higher abundance of *Bacteroides*, *Blautia* and *Clostridium*. Cluster Type 2 seemed not to be completely homogenous (with Types 2A and 2B), which initially suggested the possibility of the occurrence of a third enterotype similar to that found by Arumugam et al. [43]. Further analysis did not confirm this hypothesis and a higher abundance of *Ruminococcus* or *Blautia* in sub-cluster Type 2B was not seen. This sub-cluster seemed to be a type of mixture from clusters Type 1 and Type 2A. Due to this we took into consideration in further analyses only two enterotypes (original clusters) of bacteria. Moreover, patients with SZ were clustered at the level of KEGG genes, modules, and pathways. The severity of symptoms measured at the beginning of treatment varied, depending on the predicted metabolic activity of the bacteria. Other studies also have observed a relationship between the composition of bacteria and the severity of symptoms in SZ patients. Zheng et al. [11] demonstrated that PANSS was negatively correlated with Veillonellaceae and was positively correlated with Bacteroidaceae, Streptococcaceae, and Lachnospiraceae. Schwartz et al. [10] found greater microbial abnormalities in SZ patients than in controls. In addition, increases in the number of Lactobacillus group bacteria were positively correlated with the severity of various symptom domains in SZ patients and were negatively correlated with the global assessment of functioning. Moreover, responses to the treatment were worse in patients with severe microbiota alterations. Furthermore, Shen et al. [45], using the PICRUSt analysis, infer that vitamin B6 and fatty acid metabolic potential differed significantly between SZ patients and controls. Therefore, there are potential relationships between predicted metabolic changes and the severity of symptoms in SZ patients, as shown in Table 3. It is important to note that the PICRUSt approach using in prediction of bacterial metabolic activity should be treated with caution and followed by metagenomic analyses to explain such findings in humans. The median NSTI score was 0.11 (interquartile range of 0.05) suggesting a reasonable accuracy of the prediction, however, some closely related reference genomes were not available.


PANNS—The Positive and Negative Syndrome Scale, CGI-S—The Clinical Global Impressions Scale.

#### *J. Clin. Med.* **2019**, *8*, 1605

Our study has several strengths that should be highlighted. (1) The applied treatment resulted in expected clinical e ffects. The relationship between duration of untreated psychosis (DUP) and poor general symptomatic outcomes was confirmed, and the longer DUP was associated with more severe positive and negative symptoms. Additionally, OLZ treatment caused weight gain. This observation is in agreemen<sup>t</sup> with that in another experimental study [65]. (2) During the treatment, the patients were under the same controlled hospital conditions (diet, drug intake, and clinical monitoring), and a washout was used before treatment, thus providing a "unification" of the environmental impact on the fecal microbiota pattern. Consequently, we speculate that such conditions diminish the impact of common environmental factors that permanently shape gu<sup>t</sup> microbiota composition and underline the association between the disease and treatment. (3) Weight gain at the beginning of OLZ treatment is very important because it determines the further development of cardiometabolic risk factors [22,23,66]. (4) Although the study group was not homogeneous, the symptoms were observed every day in our psychiatric clinic. Patients were previously treated with other pharmaceuticals, which might have affected the microbiota composition. Such situations might lead to resistance against psychotropic drugs, probably leading to no impact of OLZ on the microbiota.

There are certain limitations of our study that need to be discussed. First, the sample size was small and heterogeneous (drug-naive and previously-treated patients). No formal sample size calculations were employed for this analysis, but the cohort size was based on what was previously su fficient to test microbiotic changes in schizophrenia patients [67], and/or the influence of antibiotics [68] and risperidone administration [69] on gu<sup>t</sup> microbiota. This limitation should be attributed to rigorous inclusion and exclusion criteria as well as the short duration of the whole study (17 months). Thus, studies with a greater sample size are needed to further examine the associations between OLZ treatment and gu<sup>t</sup> microbiota structure. Second, the composition of intestinal bacteria varied among individuals, and inter-individual variation within the gu<sup>t</sup> ecosystem of patients was high. Third, in individual studies (also experimental), various taxonomic groups of bacteria were analyzed only in stools. The composition of bacteria in feces is more stable and is not influenced by external factors compared with the composition of bacteria in the small intestine. Changes in the microbiota of the small bowel have a much greater e ffect on the metabolic functions of the human body. Therefore, further experimental studies should pay more attention to this issue [70,71], although an invasive way of sampling intestine biological material remains di fficult and holds several ethical concerns. Fourth, there was a lack of long-term follow-up, which is especially important in case of metabolic consequences of OLZ treatment. Fifth, we did not compare the results between SZ patients and healthy subjects or patients receiving placebos. Matched controls with similar lifestyle should be used to exclude false-positive results. However, the general lifestyle in patients diagnosed with SZ was found to be divergent from that observed in healthy people [72]. Therefore, a placebo approach was impossible mainly due to ethical and organizational concerns. Sixth, changes in dietary and living conditions during the hospital stay might be another limitation of our study. However, enterotypes *Prevotella* and *Bacteroides* are strongly associated with long-term diet. It was shown that microbiome composition changed detectably within 24 h of initiating a high-fat/low-fiber or low-fat/high-fiber diet, but that enterotype identity remained stable during the 10-day study [73]. Therefore, a change of diet after admission to hospital should not a ffect W0 microbiota. After this all patients received the same diet, and it seems that this factor should also not significantly a ffect the influence of OLZ on W6 microbiota composition.
