Introduction

Tuberculosis (TB) is an infectious disease caused by Mycobacterium tuberculosis, currently having one of the highest mortality rates [1,2,3] and causing the death of approximately 1.3 million people worldwide [4,5,6]. In the last decades, a growing issue concerning this infectious disease is the increasing number of cases resistant to first-line antibiotics (i.e., rifampicin and isoniazid), which also represents higher costs for health services and a higher burden of disease [7, 8]. As a solution to the treatment of TB-resistant patients, evidence-based guidelines and world medical societies contemplated different treatment options, including extended-spectrum antibiotics with robust effectiveness, such as cycloserine (d-4-amino-3-isoxazolidine) [9,10,11]. The latter competitively inhibits two essential enzymes for the synthesis of the cell wall of Mycobacterium tuberculosis, exerting its antibiotic effect [12, 13]. However, the therapy with cycloserine (CS) also correlates with the appearance of psychotic symptoms, since CS can penetrate the brain–blood barrier, inhibits GABA-transferase, and interacts with N-methyl-d-aspartate (NMDA) receptors in the central nervous system [14,15,16]. Although CS may induce psychosis during treatment, its appearance is considered a non-frequent condition among drug-resistant (DR) TB patients [17,18,19]. For instance, a longitudinal study reported that in a group of 144 patients with multidrug-resistant (MDR-TB) treated with CS, four patients (8%) developed psychosis [12].

Moreover, a pharmacological safety and surveillance study estimated that the global combined prevalence of adverse drug reactions due to CS was 9%, and 5.7% for psychiatric disorders [17]. In addition, of the 445 safety reports of cases reported since 1970 attributed to CS, 75% were attributed to neuropsychiatric conditions, and of these, 14% related to psychosis [17], considered in this study also a not very frequent condition. Since the appearance of induced psychosis during a CS treatment is very low, most published studies concerning cycloserine-induced psychosis are restricted to case reports. Although case reports are unsuitable for inferring causality and cannot be extrapolated, they often provide essential aspects missing in population studies, especially in rare conditions [20], such as cycloserine-induced psychosis. In this sense, conducting a systematic review of case reports would help summarize clinical and therapeutical aspects of CS-induced psychosis in MDR-TB over time [20,21,22,23]. Therefore, the main objective of this study is to perform a systematic review to describe the main clinical characteristics, comorbidities, therapy, and clinical outcomes of the cases reported in the literature on cycloserine-induced psychosis in DR-TB adult patients.

Materials and methods

Study selection criteria

First, we included studies that met the following inclusion criteria: (1) case reports of patients over 18 years; (2) patients with the diagnosis of DR-TB (MDR-TB or extreme drug-resistant TB-XDR-TB-); (3) patients with DR-TB, which have developed psychosis due to the administration of CS; (4) case report articles, case series and letters to the editor describing TB patients with CS-induced psychosis including at least the following criteria: socioeconomic data, clinical presentation and description of the psychosis, course of the illness, psychosis treatment, and clinical outcomes of the psychosis treatment.

In addition, we excluded studies if the reported participants were under 18 years, did not have TB, or were not in treatment with CS. Furthermore, we excluded studies with patients with a previous history of schizophrenia spectrum disorders or bipolar disorder, with illegal substance misuse during CS treatment, and with a current disease that could explain the induced psychosis (e.g., cerebral tumor or stroke) during the CS treatment. Studies that were not case report articles, case series articles, or manuscripts describing TB patients with CS-induced psychosis were excluded. Finally, articles that were not available as full-text were also excluded.

Search strategies

First, a systematic literature search was conducted between 07/23/2022 and 08/24/2022 in different scientific literature databases, including MEDLINE (PubMed, National Center for Biotechnology Information, National Library of Medicine, United States of America), EBSCO (EBSCO Industries Incorporate, United States of America), Web of Science (Clarivate Analytics, United States of America), CENTRAL (Cochrane Central Register of Controlled Trials, Cochrane Library, United States of America), SciELO (Scientific Electronic Library Online, Brazil), and Google Scholar (Google Incorporate, United States of America). In each case, the literature search was carried out by combining different Boolean operators (e.g., “psychosis”, “tuberculosis”), having the word “cycloserine” in the operator. More details concerning the systematic literature search, combination of Booleans, and the Boolean formulae are described in Additional file 1.

Finally, between 10/03/2022 and 10/05/2022, the search for articles was conducted again in the databases mentioned above using the same search strategy. We found 21 additional articles (3437 articles, Additional file 1); however, the additional articles were not relevant publications to this systematic review or they not fulfill the inclusion criteria, not affecting the results that we found, concluding the search on 10/05/2022.

Selection process

Posteriorly, the literature search results were extracted as .csv, .ris, or .txt files, depending mainly on the search platform and database used. In the case of the search platform Google Scholar, we used the free access software Publish or Perish [24, 25] to extract the results of the systematic literature search in the formats mentioned above. After extraction, all obtained data files were imported into the free access software Zotero v.6.09 (Corporation for Digital Scholarship, United States of America) to remove duplicated articles or records.

Afterward, we exported the remaining data without duplicates in the web application Rayyan.ai [26], where the authors ACS, SVQ, and LUE independently selected all case reports, case series, and correspondence using the title or abstract related to cycloserine-induced psychosis. Disagreements in the articles’ selection were resolved by consensus among all the authors. Later, each author performed a screening of the articles to exclude all those articles that did not meet the inclusion criteria. Likewise, the disagreements concerning inclusion and exclusion criteria were resolved under the consensus modality. Neither the year of publication of the articles nor their language was an exclusion criterion. In case of possible language issues, we used a free access translator (Google Translate, Google Inc., United States of America) to obtain the main information of the manuscript.

Quality of studies

The evaluation of the risk of bias and the quality assessment was carried out using The Joanna Briggs Institute (JBI) Critical Appraisal tool for use in Systematic Reviews [27], which was designed to evaluate the methodological quality and the risk of bias in case report studies [28,29,30,31].

The remaining articles were randomly assigned to three authors (ACS, SVQ, and LUE). During the quality assessment, each author could not reveal to the other colleagues the title or content of randomly assigned articles to make the quality assessment procedure as blind as possible. Authors were also asked not to ask for help or support from other authors while reading and assessing the quality of the assigned case reports.

If the articles did not meet the criteria for a low risk of bias described in the JBI tool, the article was sent to a second reviewer for evaluation before exclusion. In this case, the second reviewer received the article without any previous assessment or judgment of the first one to avoid bias at the moment of the second quality assessment. Finally, a third reviewer determined the article’s inclusion or exclusion in case of discrepancies between the first and the second evaluators. For this purpose, we established a structure for the quality assessment in the following order (first–second–third reviewer): ACS–SVQ–LUE, LUE–ACS–SVQ, SVQ–LUE–ACS.

Naranjo adverse reaction probability scale

A frequent limitation of case reports is to establish causality in adverse drug reactions, and this is due to different factors, such as the subjectivity of the case report and specific characteristics of the patient. One way to deal with these limitations and standardize the information presented is through the Naranjo algorithm (Naranjo Adverse Reaction Probability Scale) [32]. This tool is widely used in case reports and consists of 10 criteria with which a score is assigned that determines one of the four categories: doubtful (0 points), possible (1 to 4 points), probable (5 to 8 points), and definite (≥ 9 points) [33].

If the included case report article did not report the scores following the Naranjo algorithm, we used the information described in the case reports to estimate the category corresponding to the Naranjo algorithm, as also recommended in other studies [34].

Data extraction

Concerning the data extraction, three authors (ACS, SVQ, LUE) independently carried out this process using a template programmed for the study in Microsoft Excel. Again, all included articles were randomly assigned among these three authors, collecting and registering information concerning (1) the year of publication, (2) patients’ sociodemographic characteristics (e.g., sex, age, and country), (3) clinical characteristics of the CS-induced psychosis, (4) treatment used against the psychotic episode (e.g., antipsychotic treatment, CS treatment discontinuation, CS dose reduction or maintenance), and (4) clinical outcomes of the CS-induced psychosis.

Statistical analysis

Since this systematic review sought mainly to describe the characteristics of CS-induced psychosis in the literature, we applied descriptive statistics to present the most frequent symptoms of CS-induced psychosis and the sociodemographic characteristics and treatment used. For this purpose, we presented the quantitative data using the appropriate measures of central tendency (mean with standard deviation and median with interquartile range). Qualitative data, mostly dichotomous, were expressed using percentages and frequencies. If needed, data were presented in tables and graphs for better readability. Concerning data analysis, SPSS software (International Business Machines Corporation, New York, United States of America), version 26.0, was used for the descriptive data.

Results

General characteristics of the included reports and patients

Figure 1 shows the flow diagram according to the PRISMA protocol for the studies included in this systematic review of case reports. Initially, 3416 articles were identified; 1837 duplicates were eliminated, and 1579 publications remained. Subsequently, 1531 articles were removed after reviewing the titles and abstracts because of irrelevance to the study objective. Of the remaining 48 articles, 24 could not be retrieved and were excluded, and 3 articles were excluded, since the reported patients were underage. Finally, 21 articles were assessed with the JBI tool after the screening. One article [35] was excluded from the quality assessment process due to the low quality of information presented and the high risk of bias.

Fig. 1
figure 1

PRISMA 2020 flow diagram of systematic search for case reports concerning CS-induced psychosis and tuberculosis. Flow diagram sample extracted from: Page et al. [63]

Full size image

A total of 20 articles [36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55] met the inclusion criteria and survived the quality process, encompassing 22 patients from eight different countries (Table 1) and most patients from India (10 patients, 45.45% of the cases included). One of the articles was published in Spanish [53] and another in Japanese [37], while the rest were published in English.

Table 1 Included case report studies and their characteristics
Full size table

Of the 22 patients included in the case reports, 15 (68.18%) were male, and 7 (31.82%) were female patients. In addition, the mean age (standard deviation) among the participants was 31.45 (10.88) years.

Regarding past medical history, 15 (68.18%) had no medical conditions. However, 2 (9.09%) patients had diabetes mellitus as comorbidity, 1 (4.54%) had hypothyroidism, 1 (4.54%) had an episode of meningeal tuberculosis in the past, and 1 (4.54%) had suicide attempts during TB therapy. In this patient’s case, the study authors reported no relevant psychiatric history. Of the included studies, one patient had a family history of psychiatric disorders (major depression and bipolar disorder), and one patient had a family history of alcoholism; however, there were no reports of medical comorbidities for both patients. Finally, 19 (86.36%) reported having no illegal drug, alcohol, or nicotine dependency. However, 3 (13.64%) participants reported having used alcohol regularly in the past, and one (4.54%) patient reported the last use of cannabis and cocaine 9 weeks before admission. No reported patient consumed concomitantly illegal drugs or alcohol during the treatment with CS.

Clinical characteristics of DR-TB patients and CS-induced psychosis

Concerning the type of DR-TB patients, 21 of 22 patients (95.45%) had multidrug-resistant TB (MDR-TB), and one patient (4.54%) had renal tuberculosis (Table 1). Moreover, there were no patients with extreme drug-resistant TB (XDR-TB). Concerning the CS therapy, the mean dose (standard deviation) of CS given as treatment was 631.58 (174.17) mg/die, and the mean duration of CS treatment (standard deviation) until the onset of psychotic symptoms was 169.09 (239.52) days. Within the antitubercular treatment, the reported patients included in this systematic review received mostly drugs, such as pyrazinamide (14 patients, 63.6%), levofloxacin (12 patients, 54.5%), ethionamide (12 patients, 54.5%), kanamycin (11 patients, 50%) and pyridoxine (10 patients, 45.5%) (Table 1).

Table 1 also mentions the details of the clinical presentations of the patients in the included case reports with MDR-TB and cycloserine-induced psychosis. Regarding the duration of psychosis, the median duration of psychosis during treatment with cycloserine was 13 days (39–4 days; interquartile range: 35). The most common symptoms reported in patients with MDR-TB and cycloserine-induced psychosis were delusions (15 patients; 68.2%), followed by aggressiveness (15 patients; 68.2%), insomnia/decreased sleep (13 patients; 59.1%), hallucinations (12 patients; 54.5%), irritability (10 patients; 45.5%) and incoherent or disorganized speech (10 patients; 45.5%). Concerning the aggression observed in the case reports, 14 of 22 reported patients presented aggressive behavior, mostly against other persons, while one of the 22 included patients presented self-aggression (Table 2). In this last case, the patient died as a result of suicide due to exsanguination as a cause of self-injury cuts in the peripheral vessels.

Table 2 Main characteristics of psychiatric symptoms in patients with cycloserine-associated psychosis
Full size table

In the case of the registry of the type of hallucinations, there are records of 12 (54.5%) patients reported in the included clinical cases. Of these, two (9.1%) of the patients presented visual hallucinations, one (4.54%) presented auditory hallucinations, five (22.7%) a combination of both, and in four cases (18.2%), the hallucination type was not specified (Table 2).

Concerning the type of delusions, 7 of 9 patients (77.8%) reported persecutory delusions, one of 9 patients (11.11%) reported delusions of grandeur and 1 of 9 patients (11.11%) reported jealous delusions. Finally, 6 patients reported two delusional symptoms (Table 2). Of them, four of 6 patients (66.67%) reference and persecutory delusions, and two of 6 patients (33.33%) had persecutory and delusions of grandeur (Table 2).

Clinical treatment of MDR-TB patients with CS-induced psychosis

Treatment of cycloserine-induced psychosis varied between reported patients. Of the 22 patients reported, only 4 (18.2%) did not receive antipsychotic treatment. Of the remaining 18 (81.8%) patients reported, the majority (10 patients; 45.5%) received monotherapy with one antipsychotic (2 participants received risperidone, four olanzapine, two haloperidol, 1 received quetiapine, and 1 received chlorpromazine). Of the patients who received a combination of 2 antipsychotics (4 patients, 18.2%), two received a combination with olanzapine and haloperidol, 1 received chlorpromazine and haloperidol, and one patient received olanzapine and quetiapine. Finally, four patients (18.2%) received a combination of three antipsychotics: 2 of them received chlorpromazine + risperidone + haloperidol, 1 of them received haloperidol + promethazine + olanzapine, and 1 of them received haloperidol + chlorpromazine + olanzapine.

Of the adjuvant treatments to antipsychotic therapy, six (27.2%) patients received benzodiazepines (lorazepam, nitrazepam, and clonazepam), 3 (13.6%) patients received anticholinergic agents (benzhexol), and 1 (4.54%) received valproic acid.

Parallel to psychopharmacological therapy, 11 (50%) of the reported patients stopped receiving CS, while 2 (9.19%) continued CS treatment, and 2 (9.19%) reported patients the dose of CS was reduced (Table 1). In addition, two reported patients (9.19%) had their cycloserine dose wholly suspended for a defined period (Table 1). Finally, in 3 (13.6%) reported cases, the antitubercular therapy (including CS) was withheld (Table 1).

Finally, 20 of the 22 reported patients (90.9%) presented a clinical improvement of psychotic symptoms with the reported therapeutic strategies. On the other hand, only one patient reported partial symptom improvement with low-grade labile mood and psychosis (Table 1). Finally, in the case of one patient, suicide was committed by exsanguination by cutting himself in different regions of the body and did not present an improvement in psychotic symptoms despite treatment (Table 1).

Quality assessment of case reports: CS-induced psychosis in DR-TB patients

In general, the quality of the case reports was good, showing that the vast majority of the articles had a low risk of bias (Additional file 2). In this sense, it was determined that 17 (85%) of the reports presented a very low risk, while 3 (15%) presented a low risk. A total of 3 (15%) studies did not report the number of doses of cycloserine used in patients with MDR-TB. On the other hand, in the case of the record of symptoms, of the reported cases that described the presence of hallucinations, 4 (20%) studies did not specify what type of hallucinations they were (e.g., visual or auditory). Finally, only one study described the use of cycloserine for treating MDR-TB after using other antibiotics. At the same time, the rest of the reported cases indicate a concomitant use of cycloserine with other antibiotics. The quality assessment of the individual studies is in detail in Additional file 2.

In the case of the probability of the reaction to an adverse event, in this case, psychosis induced by cycloserine, a total of 6 articles used the Naranjo algorithm in patients who use cycloserine. In case this algorithm had not been described in the case reports, the score was established according to the information presented in the article. Of the reported articles, 18 showed a “probable” Naranjo index for cycloserine-induced psychosis (5–8 points), while the rest scored as “possible” (1–4 points).

Discussion

The following systematic review of case reports identified 20 articles (years of publication between 1965 and 2022) concerning CS-induced psychosis in DR-TB patients (mean age: 31.45 years), whose majority did not present comorbidities. Furthermore, the median duration of the CS-induced psychosis was 13 days, and the most prevalent psychotic symptom among the patients was delusions (mostly persecutory). Concerning the antipsychotic treatment against CS-induced psychosis, it was found that most received at least one antipsychotic, and parallelly CS treatment was stopped. Of the reports included in this review, 20 of 22 patients showed a clinical improvement. However, only one reported patient (treatment: risperidone and clonazepam) committed suicide by inflicting himself on multiple cuts and, therefore, dying of exsanguination. Finally, most of the included case reports showed a very low risk of bias at the time of evaluation; also, the Naranjo index showed psychosis as a “probable” adverse reaction of CS in most of the reports.

Regarding the sociodemographic data, there are similarities between our findings and previous studies with CS-induced psychosis. For example, a retrospective study reported that younger ages represented a risk factor for CS-induced psychosis [56]. In another observational study, a mean age of 35.7 years was reported in a sample of 144 patients, as well as a predominance of males in the sample, similar to the sample characteristics of our study [12].

To the best of our knowledge, this is the first systematic review of case reports which describes the psychopathology of CS-induced psychosis in reported DR-TB patients. Our results showed that delusions of persecution are the most frequent psychotic symptoms among the reported cases. Other studies with experimental designs, for instance that used CS as adjuvant treatment for 8 weeks in schizophrenic patients, also demonstrated a frequency increase in delusions, grandiosity, and hostility [57]. Finally, similar results are reported in an experimental study that found an exacerbation of psychotic symptoms through CS in a small group of patients with schizophrenia (hallucinations, an acute persecutory delusion, psychomotor agitation, thought disorganization, and catatonic symptoms) or caused a confusional psychosis (mainly including an obnubilation of consciousness and formal thought disorders with speech acceleration, that occur especially in intoxications) with circadian disorders [58].

Currently, some studies support the role of CS as a partial agonist of NMDA receptors binding to glycine sites [57, 59]. However, at high doses, CS can act as NMDA receptor (NMDAR) antagonists [60, 61], generating or worsening psychotic symptoms, as reported, for instance, in patients with schizophrenia [57, 62]. In addition, the dose-dependent NMDA antagonism could also explain the induced psychosis of the reported patients with DR-TB, receiving CS doses between 250 and 1000 mg. However, an observational study found that CS concentrations and the area under the curve were not associated with the appearance of psychotic symptoms in patients with MDR-TB [12], remaining unclear the pharmacological mechanisms of CS psychosis induction in patients with DR-TB.

Regarding the treatment of CS-induced psychosis, most reported patients presented an improvement with the administration of antipsychotic agents and the discontinuation of CS. This characteristic varies according to the literature reports, since a retrospective study shows the improvement of psychosis in most patients by reducing or temporarily suspending the CS dose [56]. However, the results of both studies agree that a reduction in CS exposure and the administration of antipsychotics reduces the frequency of the appearance of psychotic symptoms produced by CS.

Finally, the Naranjo adverse reaction scale showed that the induction of psychosis due to CS is probable (5 to 8 points). Some case reports presented values on the Naranjo scale (6 articles); however, the remaining reports did not mention a score in the Naranjo algorithm, which the authors of this study finally calculated. Naranjo’s tool may have some disadvantages, among which is the variability of the numerical score due to the evaluator’s opinion. In addition, the lack of information described in the article to qualify certain criteria could make it difficult to assess the probability of an adverse reaction in the case, for example, of external evaluators who were not involved in the treatment of the patient. This difficulty in reproducibility has been analyzed in other studies by comparing the score assigned by the same authors of the case report and that obtained by evaluators only with the information provided by the article [34]. In this case, it is observed that despite variation in the numerical score, this did not influence a significant change in the assigned category [34].

The main strength of the present study includes a description, through a systematic search, of psychotic symptoms in patients who use CS due to DR-TB, since psychosis due to CS is less frequent (between 8 and 14%) [12, 17].

However, it is important to mention that this study has limitations that must be considered. First, the small sample size of this systematic review. Second, some reports presented incomplete data (e.g., the characterization of hallucinations or the dose of CS). In addition, five articles did not present auxiliary tests (routine laboratories, diagnostic imaging, electrophysiology, etc.). Regarding the study’s methodology, this systematic review of case reports is based on non-systematized clinical information, which influenced the presentation of the details and the quality of the articles included, being a limitation of the study. Likewise, this study is not appropriate to determine the causality of the psychotic event due to the non-randomized and anecdotal methodology of the clinical cases. Finally, the accessibility of some articles was very restricted, despite contacting the corresponding authors on multiple occasions, in such a way that in these cases, only the abstract of the case report or the title of the report was available.

Conclusions

In light of our results, we conclude that the most frequent psychotic symptom of CS-induced psychosis was persecutory delusions. In addition, CS-induced psychosis is of short duration that shows mostly a symptom improvement after medical treatment, involving the CS withhold/suspension and antipsychotic treatment. Future studies should clarify possible associations between psychotic symptoms and CS serum concentrations and observe the risk factors associated with the development of psychosis due to CS in patients with DR-TB.

In clinical practice, it is important for practitioners to be aware of the potential adverse effect of cycloserine (psychosis) and conduct thorough patient evaluations during admission and follow-up rounds. If there are symptoms that suggest acute psychotic disorders, it is advisable to consult with a hospital psychiatrist, such as a liaison psychiatry service. In cases where a patient may be developing psychosis, coordination with a hospital psychiatrist is crucial to explore possible organic causes of psychosis and conduct a comprehensive third-party medical history to assess if there were similar psychotic symptoms in the past. The diagnosis of a cycloserine adverse effect should be considered if the induced psychosis is causally related and other causes are ruled out. Unfortunatelly, there are no biomarkers to indicate drug-induced psychosis, so close clinical examination and follow-up of the patient is essential in such cases.