Myxedema Psychosis: Systematic Review and Pooled Analysis

Background

Hypothyroidism is a common disease with an estimated global prevalence of 0.1–3.6%.^1–4 The Committee on Myxedema of the Clinical Society of London issued the first report that described the development of delusions and hallucinations in almost half of hypothyroid patients (109 patients).⁵ Sixty years later, in 1949, Asher et al reexamined this relationship in fourteen patients who had psychosis and clinical evidence of hypothyroidism. The patients received thyroid hormones supplements, with nine patients achieving full recovery.⁶ He labeled this association “myxedema madness,” which later was renamed “myxedema psychosis” (MP).⁶ MP is a secondary psychotic disorder resulting from other medical conditions according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5).⁷

The underlying pathophysiology is poorly understood. Previous research linked hypothyroidism to changes in neurometabolic activity that might contribute to MP, including;⁸ tyrosine hydroxylase imbalance in the anterior locus coeruleus;⁹ abundance of T3 receptors in the amygdala and the hippocampus;¹⁰ altered serotonin-mediated neurotransmission^11,12 and attenuation of cerebral regional blood flow and glucose metabolism.^13,14

Diagnostic discrimination between MP and other secondary psychoses is clinically relevant as the management differs according to the exact etiology. An important differential diagnosis of psychosis in hypothyroidism patients is Hashimotos’ encephalopathy (HE), also called steroids responsive encephalopathy with autoimmune thyroiditis.¹⁵ While the pathophysiological mechanism underlying MP is related to brain neurochemical alterations accompanying thyroid hormones deficiency, neuropsychiatric changes in HE are caused by an autoimmune response not directly linked to hypothyroidism. This explains the excellent response to steroids in most HE cases.¹⁵

Seventy years have elapsed since Asher’s description, yet, little is known about MP, likely due to the paucity of available literature.⁸ Thus, we aimed to review the literature and synthesize data on its clinical symptomatology, diagnosis, management strategies, and clinical outcomes.

Methods

This systematic review complied with preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines.¹⁶ The review protocol was registered at PROSPERO (registration number: CRD42020160310) and was published.¹⁷

Eligibility Criteria

Observational studies, case series, and case reports providing data on patients diagnosed with MP were eligible for inclusion. We only included reports on adult patients (18 years or older) with confirmed hypothyroidism (thyroid stimulating hormone> normal range plus low thyroid hormones, or clinical evidence of hypothyroidism) and psychotic features meeting the DSM-5 criteria of psychosis due to a general medical condition in whom myxedema psychosis was the likely diagnosis as per the treating physician. We excluded cases of Hashimoto’s encephalopathy (HE), thyroxine-induced mania, subclinical hypothyroidism, secondary hypothyroidism (due to possible confounding by other hormonal imbalances or mass effect), or cases with alternative diagnoses more likely than MP.

Search Strategy and Information Source

We conducted a comprehensive search in the following databases; PubMed, Medline, EMBASE, Google Scholar (first 300 hits), and Cochrane databases for studies published from 1/1/1980 to 31/12/2019. We included articles labeled “letter to the editor” if they were, in fact, case reports. We limited our search to articles written in the English language only. We used combinations of free text, keywords, Emtree, and MesH-terms, including psychosis, psycho, madness, psychiatric, hypothyroid, myxedema, myxoedema. Search strings used in each database are detailed in the supplement. (Supplementary 1) We also performed a snowball search in bibliographies of identified full‐text articles and relevant review articles.

Screening, Data Extraction, and Quality Assessment

Two independent reviewers (MFHM) and (SS) performed the literature search and screening. First, the titles and abstracts were screened. Subsequently, the full text of potentially eligible articles was reviewed and assessed for inclusion. At each step, the two reviewers discussed discrepancies noted, and if consensus could not be reached, a third reviewer (MD) settled the discrepancy per protocol. We used a web-based literature screening application (Rayvan; http://rayyan.qcri.org) to conduct article screening and duplicate removals.¹⁸

We extracted general data on included publications such as type, author, year, and journal as well as demographic data of the patients reported such as sex, age, gender, history of psychosis, history of hypothyroidism, and causes of hypothyroidism. Moreover, data on clinical presentation data was extracted, including psychiatric presentation, duration of psychosis, hypothyroidism symptoms or signs, associated rhabdomyolysis, cranial imaging finding; electroencephalography (EEG) findings; thyroid stimulating hormone level; thyroid hormone levels; anti-thyroid peroxidase status; creatinine kinase levels. We used the tool proposed by Murad et al to adjudicate the quality of included case reports and series.¹⁹ The tool comprises eight questions assessing four domains (selection, ascertainment, causality, and reporting). We generated an overall score, and we then graded the quality as either good (> 5), fair (4–5), or poor (< 3).

Statistical Analysis

We used the Jamovi 1.1.9 software for statistical analysis.²⁰ Descriptive statistics were applied to summarize data using the median (IQR) for continuous variables and frequencies for categorical variables. We used (n/N) and percentage values for presenting numbers of cases with a specific characteristic amongst cases that reported either the presence or the absence of this characteristic. Acknowledging the subjective nature of reporting in case studies, the two reviewers had to agree on whether a specific characteristic was present in any given case before inclusion in the final analysis. Exploratory multivariate logistic regression including potentially clinically relevant variables (gender, age, symptoms duration, TSH level, FreeT4, antipsychotic drugs duration, IV thyroid hormone therapy, and starting thyroxine dose) associated with recovery (resolution of psychosis) or rapid recovery (less than 2 weeks) was also performed.

Results

The initial search retrieved 2733 articles; 50 additional articles were identified through other means, of which 71 references describing 75 cases were included for the final analysis.^21–91 The PRISMA flow diagram is shown in Figure 1. All included studies were case reports due to the absence of other forms of evidence (Table 1). Quality assessment utilizing the methodological quality and synthesis of case series and case reports tool revealed fair to good quality of most of the included cases (Supplement 2).

Table 1 Summary of Included Case Studies

Figure 1 PRISMA flow diagram.Note: PRISMA figure adapted from Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. doi: 10.1136/bmj.n71. Creative Commons.93

Baseline Characteristics

The female-to-male ratio was 2:1. The median age was 42 [32–56] years, with the oldest case reported aged 90 years. The majority of cases were Caucasians, 44%, followed by Asians, 36%. 53% of patients had no prior history of hypothyroidism, and 82% had no prior psychosis history. Autoimmune thyroiditis was the most common reported cause of hypothyroidism 51%.

Clinical Features

The median duration of psychotic symptoms was 14.5 [7–82.5] days ranging from two days to three years. Delusions were present in 91%. The most common form of delusions was paranoid/persecutory 84%. Hallucinations were present in 77.5%, with auditory hallucinations being the most prevalent 77.6%. Manic symptoms accompanied psychosis more than depressive symptoms, 52% and 36%, respectively (Table 2). Hypothyroidism symptoms and signs were not always reported, and when presented, often lacking sufficient details. Hypothyroidism symptoms were present in 63% (26/41) of the cases. Only 22 cases described the nature of hypothyroidism symptoms. Fatigue occurred in 63% (14/22), weight gain 36%, cold intolerance 36% (7/22), and hoarse voice was seen in 18% of the cases (4/22). Hypothyroidism signs were present in 75% of the cases (39/52). The most common abnormal findings were dry skin 60% (23/43), facial or pretibial edema 52% (20/43), delayed relaxation, or diminished deep tendon reflexes (DTR) 47% (18/38), while hoarseness of voice occurred 26% (10/38).

Table 2 Summary of Baseline Characteristics, Clinical Features, and Diagnostic Workup of Included Cases

Laboratory Testing

The median thyroid-stimulating hormone median value was 93 mIU/L [60–93]. The median-free T4 was 0.2 ng/dl [0.13–0.39]. The median thyroid peroxidase value was 138 IU/L [82.5–323]. In 7 of the reported cases, creatinine kinase was observed abnormal with a median value of 4490 [1767–9485] IU/L. Lumbar puncture was reported in six patients. Analysis of cerebrospinal fluid revealed mild protein elation in two cases (33%) and was found normal in four cases (67%).

Diagnostic Imaging

Cranial magnetic resonance imaging was normal in 75% (12/16) of the cases and showed structural brain changes in four patients, including crescent-shaped foci of T2 hyperintensity visualized as slight effusion below the dura matter (n=1), nonspecific white matter changes (n=1), and age-related atrophic changes (n=1). Similarly, patients who underwent cranial computed tomography displayed normal brain scans in 89% of cases (24/27). The electroencephalogram was normal in 73% (11/15) and showed generalized slowing without a focal change in 27% of cases (4/15) (Supplementary 3).

Antipsychotic Medications

Antipsychotic medications were utilized in the treatment of 92% (n=55/60) of the cases. The median duration of antipsychotic use was 1.8 [0.6–8] weeks. The longest antipsychotic treatment duration was 39 weeks.

Thyroid Hormone Supplementation

The median initiating and maintenance dose of thyroxine was 100 mcg [50–141]. Intravenous thyroid hormone therapy was administered in 10% of the patients (5/50). In 85% (44/52) of the cases, no thyroxine loading was given. Triiodothyronine was administered to 8% (n=5/58) of the cases.

Steroids

Steroids were administered in 3% of the cases (2/67). The median dose used was equivalent to 50 mg of prednisolone that was used for periods of three days and two weeks, respectively.

Outcome and Follow-Up

Clinical outcome data were reported for 96% of the cases (72/75). The majority of patients, 97% (66/68), required hospitalization, and 93% demonstrated remission. However, two patients (3%) showed no improvement or residual psychosis,^23,75 while another three cases (4%) displayed recovery of psychosis with persisting residual deficits in cognition, memory, orientation, attention.^31,66 The duration-to outcome occurrence was reported in 83% (62/75) of the cases. The median duration to the outcome (recovery) was 1.93 [0.8–2] weeks. The use of intravenous thyroid hormone supplementation (4/46) was associated with faster recovery compared to oral administration (0.55 [0.5–0.85] weeks vs 2.0 [1–2.85] weeks (p= 0.022). The univariate analysis also revealed a trend towards a shorter duration of psychosis (p= 0.0502) with IV thyroid hormone therapy. However, this effect could not be confirmed in the multivariate analysis (Table 3). Recovery duration did not differ between patients who received triiodothyronine and those who did not 0.5 [0.37–1.95] weeks vs 2 [1–2] weeks, p= 0.2). In the multivariate analysis (Table 3), gender (p= 0.33), age (p=0.46), symptoms duration (p=0.98) TSH level (p=0.29), FreeT4 (p=0.32), antipsychotic drugs duration (p=0.22), IV thyroid hormone therapy, and starting thyroxine dose (0.52) were not associated with shorter duration of psychosis (<2 weeks).

Table 3 Table Summarizing the Result of the Multivariate Analysis

Discussion

The major finding of this systematic review is that evidence on the pathophysiology as well as the clinical course and management of MP is limited to case reports. Descriptive pooling of extracted data from these reports and exploratory analysis indicates that MP can manifest with a wide range of psychiatric and physical symptoms and is commonly treated with antipsychotics and thyroid hormone supplementation. Although the vast majority of patients needed to be hospitalized, very few displayed persisting residual deficits after treatment. Prospective research is urgently needed to improve our understanding of MP and identify factors that may modulate clinical outcomes in order to design standardized diagnostic and therapeutic regimens.

The prevalence of MP was not primarily studied. Therefore, it can only be indirectly estimated from studies evaluating psychotic symptoms in patients with hypothyroidism. Based on the Committee on Myxedema of the Clinical Society of London report, the prevalence of psychotic symptoms was around 50% in hypothyroid patients in the late nineteenth century,⁵ and was less than 2% in 1965 based on a study of four-hundred hypothyroid patients in which 2% of patients were described to have hypothyroidism associated mental changes (this study did not provide details about the nature of psychic changes, which could be non-psychotic or hypothyroidism unrelated).⁹²

In this review, we identified reports of MP in all adult age groups with a slightly higher proportion of reports on younger patients. Symptoms of hypothyroidism were observed in half of the cases, indicating that the absence of a prior history of hypothyroidism does not rule out MP. The majority of the cases did not have a personal nor family history of psychosis, supporting the direct link between thyroid pathology and psychosis. Although the pooled data from reported cases do not constitute a representative population, we observed a possible pattern in the clinical manifestation of the predominance of paranoid/persecutory delusions and auditory hallucinations. Manic symptoms tended to accompany MP more often than symptoms of depression. Fatigue was the most frequent somatic symptom; that is, however, nonspecific and can be seen in both primary and secondary psychiatric disorders. Although interesting, the predominance of manic symptoms and fatigue can be due to under-reporting as many clinical features were reported in few cases only. Interestingly, almost 40% of the cases did not have any physical hypothyroidism symptoms, while TSH was elevated in all cases with low thyroid hormones supporting the value of screening for thyroid dysfunction in patients with first-ever psychosis. Cranial imaging data did not suggest any relevant changes in brain structure related to MP. Almost all patients needed hospital admission. Antipsychotic medications were given to most of the patients (92%) with remission following short-term administration. Whether this indicates that termination of antipsychotic therapy after remission of psychotic symptoms is safe in patients with MP remains to be answered by prospective research.

Our pooled data of MP cases showed that administration of thyroxin was performed either intravenously or orally. Explorative comparison of synthesized data did not show the superiority of one administration route over the other, although unadjusted comparison showed a possible trend toward faster recovery after IV administration. However, this observation is solely based on cumulative case reports and needs to be viewed in conjunction with the potential increased risk of arrhythmias associated with IV thyroxin.^94,95 Administration of triiodothyronine in MP was reported small fraction of the cases and should be investigated in prospective research. In most cases, steroids were not needed, questioning the value of immunosuppressive medication in MP, contrary to myxedema coma and HE. Most cases recovered completely, and only a few cases were left with residual psychosis or cognitive deficits. The etiology of partial improvement or residual cognitive deficits could be either a chronic and irreversible metabolic effect induced by hypothyroidism, as hypothesized by Asher et al or a primary psychiatric illness precipitated by hypothyroidism. All cases that recovered improved within two weeks on average, not exceeding six weeks at most. However, not all cases had long-term follow-up details to study recurrence of psychosis in the presence or absence of dysthyroid status.

Our review excluded two reports of patients with mania that did not exert psychotic features.^96,97 Although the reported cases would be classified under the initial term “Myxedema Madness,” we have excluded them as they would not fit within the term “Myxedema psychosis.”

Prior systematic reviews examining HE,^98–100 showed that only 20–25% of patients had clinical hypothyroidism, and most of them were euthyroid. It has been stated that psychosis occurs in around a third or less of the patients (26–36%) and is usually accompanied by other features such as; seizures in up to two-thirds (59–66%) and myoclonus (36–42%). These features were not seen in our patient population and their presence may help distinguish HE from MP, especially if the patients are euthyroid. Thyroperoxidase antibodies were found positive in almost all HE cases (86–100%), while our review revealed that 50% of MP cases had autoimmune thyroiditis. This may support the diagnostic value of a negative, not a positive, thyroperoxidase antibody test to differentiate MP from HE. Elevated protein in the cerebrospinal fluid is another characteristic feature of HE occurring in 71–78%, while electroencephalogram is usually abnormal (80–98%).¹⁵ These may also help differentiate it from MP as shown in our review, abnormal CSF and or EEG are not common in MP cases. Moreover, the small number of MP cases with abnormal CSF or EEG as depicted by our review could have been simply misdiagnosed HE cases.

Our systematic review has strengths, including a comprehensive literature search following PRISMA guidelines after a priori registration and publication of the predefined review protocol. We were able to pool data from a wide variety of sources identified in the published literature. We provided data on the demographics, diagnosis, and management of myxedema psychosis. Our pooled analysis is solely based on case reports constituting a non-representative population. However, our work clearly shows a substantial research gap, substantiating an urgent need for prospective well-designed research to characterize the clinical course and characteristics of myxedema and to test tailored diagnostic and therapeutic strategies. Publication and reporting bias cannot be ruled since cases considered not interesting or those with adverse outcomes could have been under-reported. Moreover, incomplete reporting of many clinical symptoms or details’ reporting within individual cases may have biased the final conclusion with regards to background data, outcomes, and follow-up duration. Furthermore, the diagnosis of MP was a clinical diagnosis as deemed likely by the treating physicians. We limited our search to case reports starting from 1980. By doing this, we may have missed including a small number of cases; however, considering that the differential diagnosis of HE was first described in 1966,¹⁰¹ we intended to collect comparable cases in order to differentiate between both conditions. Another reason for limiting our search to the time period since 1980 was changes in presentation and symptomatology of hypothyroidism due to improved diagnosis and disease management over time.

Conclusion

Our systematic review and pooled analysis identified a substantial lack of published research on MP. Available case observations indicate that patients with MP present with a broad spectrum of psychiatric and physical symptoms lending support to the value of screening for thyroid dysfunction in patients with first-ever psychosis.