In the present study, we conducted an investigation in a cohort of patients diagnosed with both BD type I and type II to explore the connection between lithium response and a comprehensive set of variables, encompassing not only clinical factors but also temperament traits and psychopathological dimensions. Our findings revealed that among clinical factors, only the presence of a manic-depressive-interval (MDI) cycle emerged as a specific predictor of lithium response. In terms of temperament traits, both hyperthymic and irritable temperaments were identified as predictors of positive lithium response, while a history of childhood emotional and physical abuse was indicative of non-response to lithium treatment.

The observation that the MDI cycle type can be a clinical predictor of lithium response aligns with existing literature. Clinical studies conducted in the 1980s observed a higher occurrence of the MDI sequence in patients who displayed a favorable response to lithium compared to those with poor responses (Grof et al. 1987; Haag et al. 1987; Koukopoulos et al. 2013; Maj et al. 1989). Authors postulated that this phenomenon might be associated with the primacy of antimanic action of lithium during manic/hypomanic phases, as it may serve as a protective measure against subsequent episodes of depression in MDI cycles (Koukopoulos et al. 2013; Maj et al. 1989). Authors also highlighted that the course characterized by depression-mania-interval (DMI) is frequently associated with antidepressant treatment for the initial depression, often resulting in mood switches into mania or hypomania (Maj et al. 1989). Such mood switch tended to be relatively resistant to the effects of mood stabilizers.

It is worth noting that in these earlier studies the response to lithium was not consistently evaluated but rather defined based on clinical parameters, such as the absence of episode recurrence for more than a year (Koukopoulos et al. 2013) or a reduction in the frequency of episodes or hospitalizations (Haag et al. 1987). In the present study, for the first time, we have substantiated these observations using a reliable assessment tool, the Alda scale, in conjunction with a multivariate statistical model. Our results align with recent meta-analytic findings (Hui et al. 2019) that aggregated the aforementioned observations, indicating the MDI sequence cycling as a robust predictor of a positive response to lithium, with a low level of heterogeneity across the included studies.

It is interesting to highlight that recent research has largely overlooked the consideration of cycle type in assessing lithium response. This trend could be attributed to the declining use of life charts in systematically evaluating the clinical course of BD, potentially due to the time and effort required for an accurate reconstruction of the clinical trajectory of patients. Our findings underscore the importance of a lifetime approach to the treatment of patients with BD (Sani et al. 2017) and to systematically consider the type of cycle in patients with BD, in order to formulate clinical prognoses and potentially guide more effective and safer treatment strategies.

In terms of temperament traits, our findings revealed that hyperthymic and irritable temperaments are distinct predictors of a favorable response to lithium treatment. The results regarding hyperthymic temperament corroborates a prior study that demonstrated a positive correlation between lithium response and hyperthymic temperament (Rybakowski et al. 2013). Importantly, our findings confirmed this observation within a larger sample size and using a comprehensive multivariate statistical model. Clinically, these results may be indicative of lithium’s enhanced efficacy in the context of acute euphoric mania, a condition often associated with hyperthymic temperament (Perugi et al. 2001). Furthermore, hyperthymic traits tend to be linked with a predominant manic/hypomanic polarity (Azorin et al. 2015) which, in turn, is predictive of a positive lithium response phenotype (Scott et al. 2020).These findings align with the historical notion of lithium’s primary antimanic action, as first observed by Cade (Cade 1949), and are consistent with the association we identified between MDI cycle and lithium response, further supporting the relevance of the manic/hypomanic phase in predicting a favorable response to lithium treatment.

The link between irritable temperament and positive response to lithium may also be due to lithium’s antimanic effects, which counteract the intrinsic excitatory nature of irritable traits. In line, patients with irritable temperament spend more time in manic/hypomanic phases (Miola et al. 2021), are more frequently diagnosed as BD type I (Pompili et al. 2014) and present with more prodromal symptoms prior to the first manic/hypomanic episode (Zeschel et al. 2015).

It is important to acknowledge that our findings regarding irritable temperament do not match the study conducted by Rybakowski et al. (Rybakowski et al. 2013). Furthermore, they appear to contradict well-established observations that suggest a positive association between irritable temperament and mixed states (Tundo et al. 2023), which have in turn been linked to a less favorable response to lithium (Etain et al. 2017; Sportiche et al. 2017). However, it is important to emphasize that these two studies have not been consistently replicated, and recent meta-analytic data suggest that the relationship between mixed states and lithium response is inconclusive and lacks consistency (Hui et al. 2019). Additionally, lithium is increasingly recognized in clinical practice as an effective treatment for managing mixed states (Sani and Fiorillo 2020).

Our study revealed that early adverse events, specifically emotional and physical abuse, substantially predict an inadequate response to lithium treatment in patients with BD. This finding is consistent with a prior study by Etain et al. (2017), which reported an association between suboptimal response to lithium and physical abuse, but not emotional abuse. Nevertheless, the authors cautioned that the absence of association with emotional abuse may be attributed to limited statistical power. Cascino et al. (2021) is the only other study that evaluated patients using both the Alda scale and the CTQ-SF, showing similar results.

Our results show that the deleterious impact of early adverse events on treatment response may be mediated by the absence of a reduction in the overall illness activity. This aligns with the observed worsened clinical course associated with childhood trauma in BD, characterized by a higher frequency of episodes, increased hospitalizations, elevated suicidality and self-harm, compromised psychological functioning, and greater cognitive impairment (Barczyk et al. 2023; Daruy-Filho et al. 2011; Janiri et al. 2023b; Janiri et al. 2023a; Köhler-Forsberg et al. 2024). This may exert a particular influence on the clinical course of patients with an unstable temperament and high nervous reactivity, characteristics that are prevalent among most BD type II patients (Koukopoulos et al. 2006).

The substantial percentage of patients with BD reporting childhood adversities may also be linked to the relatively high prevalence of patients who do not achieve an optimal response to lithium, as reported in other studies using the Alda scale (Sportiche et al. 2017). Childhood trauma may impact various biological pathways, possibly not fully countered by lithium treatment, explaining why some patients on lithium remain at high risk for mood recurrences. Trauma may affect pathways related to inflammation, neuroplasticity, circadian systems, and premature aging (Gill et al. 2020; Janiri et al. 2017; Ridout et al. 2018). It can lead to changes in Brain-derived neurotrophic factor (BDNF) levels and inflammatory markers, potentially creating an imbalance between neuroplasticity and proinflammatory cytokines (Gill et al. 2020; Misiak et al. 2020). Molecular hypotheses related to the circadian system and telomere length are also relevant, with childhood trauma possibly contributing to poor lithium response due to long-lasting biological consequences (Ridout et al. 2018). Additionally, childhood trauma can alter brain structures, such as hippocampus and amygdala (Janiri et al. 2017), which may further impact lithium response (Sani et al. 2018).

Consistently with prior observations (Scott et al. 2017, 2020; Sportiche et al. 2017), although not confirmed by meta-analytic results (Hui et al. 2019), we also noted a link between lithium non-response and lifetime substance use. Additionally, we found a positive association between BD type I diagnosis and early age at onset with favorable lithium response. The first observation on BD subtypes aligns with previous studies (Sportiche et al. 2017) and supports the prominence of lithium’s antimanic effect. The second finding is contentious because it contradicts prior research, which has shown a positive association between lithium response and later age at onset (Hui et al. 2019; Lee et al. 2020). However, it’s worth noting that one previous study found that lithium responders were more likely to have an earlier onset compared to non-responders (Garnham et al. 2007), and another study identified age at onset between 15 and 35 years as a predictor of lithium response (Scott et al. 2020). Nonetheless, none of these associations in our study survived multivariate analyses.

Before presenting our conclusions, we need to acknowledge certain limitations that could affect the generalizability of our findings. Firstly, the cross-sectional design of our study limits our ability to establish causal relationships. However, we mitigated this limitation by employing a multivariate model to predict lithium response, while accounting for relevant confounding factors. Secondly, despite including both BD type I and type II subtypes, there is an imbalance in our sample, with a higher proportion of patients with BD type I (77%) compared to type II (23%), representing a potential limitation in terms of representativeness. Thirdly, clinical variables and treatment response were assessed retrospectively. Nevertheless, we used the Alda scale, which is considered the gold standard for evaluating lithium response (Manchia et al. 2013). Additionally, to reduce the risk of recall bias, we collected information not solely from patient reports but also from family members and close friends (who were present at least at one visit) and reviewed all available medical records. Lastly, while our study features a relatively large sample size, it is still smaller than those in recent genetic and clinical studies (Nunes et al. 2020; Scott et al. 2020). Nonetheless, this study represents the first attempt to simultaneously consider clinical variables, temperament traits psychopathological dimensions, and childhood trauma as potential predictors of lithium response.

In conclusion, our study enhances our capacity to characterize lithium response in individuals with BD by incorporating clinical variables alongside temperament traits, psychopathological traits and early adverse events. Temperament and childhood trauma, when considered alongside the type of cycling, should consistently be factored into the initial phases of BD treatment. The collective evaluation of clinical and psychopathological dimensions may prove to be more predictive of lithium response than the nosological diagnosis alone, as suggested by the current perspective of prominent researchers and clinicians worldwide (McIntyre et al. 2022). Future research endeavors could delve into the biological mechanisms underpinning the associations we have identified and their impact on lithium response.