Lifetime PMDD/PMS comorbidity in mood disorders population (mood disorder diagnosed first)

The funnel plot showed slight left-sided asymmetry (Supplementary Fig. 1), suggesting small-study effects. Egger’s test was borderline significant (−4.72, P = 0.045, one-tailed; P = 0.089, two-tailed), but Trim-and-Fill found no missing studies (Supplementary Table 6). Pooled prevalence remained at 42% (95% CI: 30%, 55%), indicating no notable publication bias. High heterogeneity (I ² = 98%) probably explains the observed asymmetry.

Lifetime mood disorders comorbidity in PMDD/PMS population (PMDD/PMS diagnosed first)

The funnel plot showed significant left-sided asymmetry (Supplementary Fig. 2), suggesting under-reporting of lower prevalence studies. Egger’s test confirmed bias (Intercept 9.59, P < 0.00003; Supplementary Table 7). Trim-and-Fill detected 2 missing studies, reducing pooled prevalence from 45% (95% CI: 26%, 65%) to 39% (95% CI: 23%, 60%; Supplementary Table 7). High heterogeneity (I ² = 99%) also contributed to variation. These findings suggest probable publication bias, but the adjusted prevalence remains within a similar range, meaning that the conclusions are not fundamentally altered.

Comorbidity between PMDD/PMS and mood disorders in concurrent diagnostic sampling

The funnel plot (Supplementary Fig. 3) and Egger’s test showed no significant asymmetry (intercept −4.02, P = 0.37; Supplementary Table 8). However, Trim-and-Fill detected 2 missing studies, slightly reducing pooled prevalence from 42% (95% CI: 25%, 61%) to 35% (95% CI: 20%, 54%; Supplementary Table 8). Adjusted confidence intervals overlap with the original, indicating only minor bias. High heterogeneity (I ² = 94%) suggests additional study variability.

Sensitivity analysis: mood disorders comorbidity in PMDD/PMS population (diagnosed first, 2-month prospective charting)

The funnel plot (Supplementary Fig. 4) and Egger’s test showed no asymmetry (Intercept −0.19, P = 0.94), and Trim-and-Fill detected no missing studies (Supplementary Table 9). Pooled prevalence remained at 49% (95% CI: 38%, 60%), suggesting no publication bias. Moderate heterogeneity (I ² = 75%) indicates that study differences are probably due to methodology rather than selective reporting.

Studies on mood disorders (mixed samples including BPDs and UDD)

Of 14 studies that included females with PMDD/PMS and mood disorder comorbidity, 1 investigated biomarkers ^{Reference Hardoy, Serra, Carta, Contu, Pisu and Biggio73} and 7 explored clinical course ^{Reference Ascher-Svanum and Miller68–Reference Christensen and Oei70,Reference Li, Tsai, Bai, Su, Chen and Liang75–Reference Payne, Klein, Zamoiski, Zandi, Bienvenu and Mackinnon77,Reference Diamond, Rubinstein, Dunner and Fieve79} alongside prevalence. Among these, 2 studies enrolled euthymic individuals with BPDs or MDD, ^{Reference Chan, Lo, Hsu, Chiu, Huang and Liao69,Reference Hardoy, Serra, Carta, Contu, Pisu and Biggio73} 2 compared depressed and euthymic individuals among those with BPDs or MDD, ^{Reference Angst, Sellaro, Merikangas and Endicott67,Reference Payne, Klein, Zamoiski, Zandi, Bienvenu and Mackinnon77} while the remaining 10 did not specify the mood state of individuals with BPDs or MDD at enrollment. ^{Reference Adewuya, Loto and Adewumi66,Reference Ascher-Svanum and Miller68,Reference Christensen and Oei70–Reference Frey, Allega, Eltayebani, Syan, Mendes-Ribeiro and Minuzzi72,Reference Hong, Park, Wang, Chang, Sohn and Jeon74–Reference Payne, Roy, Murphy-Eberenz, Weismann, Swartz and McInnis76,Reference Wittchen, Becker, Lieb and Krause78,Reference Diamond, Rubinstein, Dunner and Fieve79} Of the 14 studies, 4 also included individuals with dysthymia or persistent depressive disorder. ^{Reference Adewuya, Loto and Adewumi66,Reference Angst, Sellaro, Merikangas and Endicott67,Reference Christensen and Oei70,Reference Wittchen, Becker, Lieb and Krause78}

Across 7 studies on clinical course, females with comorbid PMDD/PMS and mood disorders exhibited higher illness burden compared to those with mood disorders alone. For instance, Payne et al ^{Reference Payne, Roy, Murphy-Eberenz, Weismann, Swartz and McInnis76} found that PMS symptoms significantly predicted the onset of postpartum or perimenopausal mood symptoms in MDD females, but not in BDI females. Payne’s subsequent study ^{Reference Payne, Klein, Zamoiski, Zandi, Bienvenu and Mackinnon77} reported that, while PMS was not familial in MDD or BPD families, MDD females with PMS had more past depressive episodes than MDD females without PMS. However, this pattern was not observed in BPDs, except that BDI females with PMS had a significantly earlier age at onset of BPDs compared to those without PMS. ^{Reference Payne, Klein, Zamoiski, Zandi, Bienvenu and Mackinnon77} Similarly, a longitudinal study found that females with PMDD who later developed mood disorders had an earlier age at onset for both BPDs and MDD, and a shorter interval between study enrollment and mood disorder diagnosis. ^{Reference Li, Tsai, Bai, Su, Chen and Liang75}

The greater illness burden associated with comorbid PMDD/PMS with mood disorders was further corroborated by findings on suicidality, metabolic health and psychiatric hospitalisations. PMDD was independently correlated with higher lifetime suicide attempt rates, even after controlling for depressive and anxiety symptoms. ^{Reference Chan, Lo, Hsu, Chiu, Huang and Liao69} Moreover, females with both mood disorders and PMS had significantly higher rates of obesity and greater overall weight than those with mood disorders alone. ^{Reference Ascher-Svanum and Miller68} Psychiatric hospitalisations were notably more frequent for females with mood disorders during the premenstrual phase compared with other menstrual-cycle phases, despite no significant differences in treatment-seeking behaviour, social impairment or premenstrual somatic/cognitive symptoms. ^{Reference Diamond, Rubinstein, Dunner and Fieve79} In a related study, females seeking PMS treatment exhibited greater premenstrual mood severity, regardless of whether PMS was prospectively confirmed, although no major differences in mood disorder history were noted between confirmed and unconfirmed PMS cases. ^{Reference Christensen and Oei70}

In terms of findings related to biomarkers, Hardoy et al ^{Reference Hardoy, Serra, Carta, Contu, Pisu and Biggio73} assessed plasma concentrations of neuroactive steroids, specifically progesterone and 3α,5α-tetrahydroprogesterone (3α,5α-THPROG), females with BPD and MDD with and without PMDD, and found no significant differences in neurosteroid levels between PMDD and non-PMDD groups after adjusting for psychiatric diagnosis.

Studies on bipolar disorders

Seven studies investigated the comorbidity between PMDD/PMS and BPDs, including 1 study on biomarkers, ^{Reference Syan, Minuzzi, Smith, Costescu, Allega and Hall107} 5 on clinical course alongside prevalence ^{Reference Choi, Baek, Noh, Kim, Choi and Ha101–Reference Liang, Yang, Liao, Yang, Lin and Wu105} and 1 focused solely on prevalence. ^{Reference Karadag, Akdeniz, Erten, Pirildar, Yucel and Polat106} Among these, 3 studies enrolled euthymic individuals with BPD ^{Reference Choi, Baek, Noh, Kim, Choi and Ha101,Reference Karadag, Akdeniz, Erten, Pirildar, Yucel and Polat106,Reference Syan, Minuzzi, Smith, Costescu, Allega and Hall107} 1 included people in different states including depression, hypo/mania, mixed and euthymia ^{Reference Liang, Yang, Liao, Yang, Lin and Wu105} and 3 did not specify the mood state of individuals with BPDs at enrollment. ^{Reference El Dahr, de Azevedo Cardoso, Syan, Caropreso, Minuzzi and Smith102–Reference Slyepchenko, Frey, Lafer, Nierenberg, Sachs and Dias104}

Of the 5 studies examining clinical correlates, all found increased illness burden in the presence of comorbidity. Slyepchenko et al ^{Reference Slyepchenko, Frey, Lafer, Nierenberg, Sachs and Dias104} reported that BPD females with PMDD had an earlier age at menarche and BPD onset, higher rates of rapid cycling and increased numbers of both lifetime and past-year depressive episodes compared to BPD females without PMDD. Another study identified a significant association between PMS and greater seasonal variation in mood, behaviour and bodily functions, with the BPD group showing a stronger association. ^{Reference Choi, Baek, Noh, Kim, Choi and Ha101} Additionally, BDII females had higher rates of PMDD/PMS than BDI females. ^{Reference Choi, Baek, Noh, Kim, Choi and Ha101} Another study found higher rates of cyclothymia and BDII in females with PMDD compared to those without PMDD, while females without PMDD were more likely to have BDI. ^{Reference Fornaro and Perugi103}

One recent study found that PMDD and BPD females had lower rates of traumatic life events but higher rates of inflammatory or autoimmune disease, sleep deprivation and family history of mental disorders compared to BPD females without PMDD. ^{Reference Liang, Yang, Liao, Yang, Lin and Wu105} Furthermore, the groups with PMS with BPD and PMDD with BPD both demonstrated significantly higher depressive symptom severity, anxiety/somatisation, cognitive impairment, psychomotor retardation, increased appetite and leaden paralysis compared with the BPD-only group. ^{Reference Liang, Yang, Liao, Yang, Lin and Wu105}

A study investigating biological rhythm disruptions found that only PMDD, only BPD and PMDD with BPD groups exhibited greater overall biological and sleep disruption compared with healthy controls, with the BPD group having greater sleep disruption than the PMDD group. ^{Reference El Dahr, de Azevedo Cardoso, Syan, Caropreso, Minuzzi and Smith102} Moreover, females with PMDD and BPD had greater social impairment and depressive symptom severity compared with the PMDD-only and healthy control groups. ^{Reference El Dahr, de Azevedo Cardoso, Syan, Caropreso, Minuzzi and Smith102}

Syan et al ^{Reference Syan, Minuzzi, Smith, Costescu, Allega and Hall107} assessed brain structure, resting-state functional connectivity and clinical characteristics in females with comorbid BPDs and PMDD, PMDD only, BPDs only and healthy controls. They observed decreased functional connectivity between the hippocampus and premotor cortex, and increased functional connectivity between the hippocampus and frontal cortex in the PMDD and BPD group compared with the BPD-only group. ^{Reference Syan, Minuzzi, Smith, Costescu, Allega and Hall107} Additionally, they noted increased grey matter volume in the caudate and decreased cortical thickness in the pericalcarine, superior parietal, middle temporal, rostral middle frontal and superior frontal cortices, but increased cortical thickness in the superior temporal cortex in the PMDD and BPD group compared with the BPD-only group. ^{Reference Syan, Minuzzi, Smith, Costescu, Allega and Hall107} PMDD and BPD females had significantly higher body mass index than the other groups, but no differences were found between the BPD-only and PMDD comorbid with BPD groups regarding several BPD severity markers, including age at onset, number of psychiatric comorbidities and psychiatric medications. ^{Reference Syan, Minuzzi, Smith, Costescu, Allega and Hall107} During the follicular phase, the PMDD and BPD group exhibited higher depressive symptom severity and disruptions in sleep, physical and social activity compared with healthy controls, with these disturbances being sustained through the premenstrual phase. ^{Reference Syan, Minuzzi, Smith, Costescu, Allega and Hall107} Despite being clinically euthymic for at least 2 months before study entry, all diagnostic groups demonstrated significantly greater subclinical depressive symptoms compared with healthy controls, with the PMDD and BPD group reporting greater severity than the BPD-only group. ^{Reference Syan, Minuzzi, Smith, Costescu, Allega and Hall107}

Studies on unipolar depressive disorders (including dysthymia)

Of the 18 studies investigating comorbid PMDD/PMS with unipolar depressive disorders (UDD), 4 examined both biomarkers and clinical course ^{Reference Klatzkin, Morrow, Light, Pedersen and Girdler83,Reference Klatzkin, Lindgren, Forneris and Girdler84,Reference Stiernman, Dubol, Comasco, Sundström-Poromaa, Boraxbekk and Johansson90,Reference Klatzkin, Morrow, Light, Pedersen and Girdler92} while 9 focused on clinical course ^{Reference Hartlage, Arduino and Gehlert82,Reference Klatzkin, Bunevicius, Forneris and Girdler85,Reference Roca, Schmidt and Rubinow87–Reference Śliwerski, Koszałkowska, Mrowicka and Szafran89,Reference Soares, Cohen, Otto and Harlow93,Reference Forrester-Knauss, Zemp Stutz, Weiss and Tschudin95–Reference Halbreich and Endicott97} alongside prevalence. Among these, 11 studies enrolled individuals who were currently euthymic with a lifetime diagnosis of MDD, ^{Reference Critchlow, Bond and Wingrove80,Reference Hartlage, Arduino and Gehlert82–Reference Klatzkin, Bunevicius, Forneris and Girdler85,Reference Roca, Schmidt and Rubinow87,Reference Cohen, Soares, Otto, Sweeney, Liberman and Harlow91–Reference Accortt, Kogan and Allen94,Reference Miyaoka, Akimoto, Ueda, Ujiie, Kametani and Uchiide96,Reference Halbreich and Endicott97} 3 compared individuals who were either currently depressed or euthymic ^{Reference Śliwerski and Koszałkowska88,Reference Śliwerski, Koszałkowska, Mrowicka and Szafran89,Reference Accortt, Kogan and Allen94} and 1 involved individuals with dysthymia/persistent depressive disorder. ^{Reference Accortt, Kogan and Allen94} The remaining 4 studies did not specify the mood state at enrollment, ^{Reference Roca, Schmidt and Rubinow87,Reference Forrester-Knauss, Zemp Stutz, Weiss and Tschudin95–Reference Halbreich and Endicott97} with 2 of these including individuals diagnosed with dysthymia. ^{Reference Roca, Schmidt and Rubinow87,Reference Miyaoka, Akimoto, Ueda, Ujiie, Kametani and Uchiide96}

Several studies highlighted the clinical impact of comorbid PMDD/PMS and UDD. Specifically, 8 studies found that comorbidity was significantly associated with increased mood symptom severity, greater frequency and longer duration of mood episodes and a higher likelihood of lifetime mood episodes. ^{Reference Hartlage, Arduino and Gehlert82,Reference Klatzkin, Lindgren, Forneris and Girdler84,Reference Klatzkin, Bunevicius, Forneris and Girdler85,Reference Roca, Schmidt and Rubinow87,Reference Śliwerski, Koszałkowska, Mrowicka and Szafran89,Reference Forrester-Knauss, Zemp Stutz, Weiss and Tschudin95–Reference Halbreich and Endicott97} For example, a prospective cohort study spanning 2 years reported that females with PMDD were significantly more likely to develop MDD (odds ratio: 14) compared to those with a history of MDD (odds ratio: 6), or with those with MDD developing PMDD (odds ratio: 3.5). ^{Reference Hartlage, Arduino and Gehlert82} In a 5- to 12-year follow-up, Roca et al ^{Reference Roca, Schmidt and Rubinow87} found that females with PMS had higher rates of subsequent psychiatric episodes, including MDD, compared to those without PMS. Similarly, Śliwerski et al ^{Reference Śliwerski, Koszałkowska, Mrowicka and Szafran89} compared four groups – PMS with MDD, MDD only, PMS only and healthy controls – and found that females with both PMS and MDD reported a higher frequency of depressive symptoms in the past week compared to those with PMS alone. Although the PMS and MDD group exhibited greater severity in negative beliefs and maladaptive metacognitions, and lower functioning in the cognitive triad, the presence of depression (with or without PMS) was the strongest predictor of cognitive distortions across the groups. ^{Reference Śliwerski, Koszałkowska, Mrowicka and Szafran89} Furthermore, 1 study found that females with both PMS and MDD were more likely to report symptoms of anger, irritability, anxiety, insomnia, hypersomnia and increased sensitivity compared to those with PMS alone. ^{Reference Śliwerski and Koszałkowska88} Another study identified a higher prevalence of current smokers among females with PMDD and a history of MDD, although this difference disappeared when combining current and past smokers in the analysis. ^{Reference Soares, Cohen, Otto and Harlow93}

Among studies focused on biomarkers, ALLO and/or P4 were evaluated in 2 studies investigating PMDD with a history of MDD. ^{Reference Klatzkin, Morrow, Light, Pedersen and Girdler83,Reference Klatzkin, Morrow, Light, Pedersen and Girdler92} Klatzkin et al ^{Reference Klatzkin, Morrow, Light, Pedersen and Girdler83} found blunted ALLO response to stress in PMDD females with an MDD history compared to those without. Furthermore, ALLO levels at baseline and blunted ALLO reactivity to stress predicted more severe symptoms in females with PMDD and a history of MDD. ^{Reference Klatzkin, Morrow, Light, Pedersen and Girdler83} Another study found ALLO levels to be significantly higher in the PMDD-only group than in the comorbid and healthy controls groups. ^{Reference Klatzkin, Morrow, Light, Pedersen and Girdler92} They also found an interaction effect such that the comorbid group had a reduced ALLO/P4 ratio compared to females with PMDD only, while the same effect was not observed in females without PMDD who had prior depression. ^{Reference Klatzkin, Morrow, Light, Pedersen and Girdler92} One study investigated pain-related phenotypes in females with PMDD and prior depression, and found that the comorbid group had significantly lower cortisol levels but higher scores of unpleasant pain compared to females without PMDD. ^{Reference Klatzkin, Lindgren, Forneris and Girdler84} In addition, a recent study examined brain activity in response to emotional stimuli in PMDD females and healthy controls, where both groups included females with a psychiatric history of depression. ^{Reference Stiernman, Dubol, Comasco, Sundström-Poromaa, Boraxbekk and Johansson90} Findings showed that females with PMDD had increased brain activity in emotion-processing brain networks, including the insula and frontal and cingulate cortices, during the premenstrual phase compared with controls, and the results were unaffected by history of depression among participants. ^{Reference Stiernman, Dubol, Comasco, Sundström-Poromaa, Boraxbekk and Johansson90}

Across studies investigating clinical course, the co-occurrence of PMDD/PMS with mood disorders was consistently associated with markers of increased illness burden and poor prognosis. ^{Reference Chan, Lo, Hsu, Chiu, Huang and Liao69,Reference Li, Tsai, Bai, Su, Chen and Liang75–Reference Payne, Klein, Zamoiski, Zandi, Bienvenu and Mackinnon77,Reference Hartlage, Arduino and Gehlert82,Reference Klatzkin, Lindgren, Forneris and Girdler84,Reference Klatzkin, Bunevicius, Forneris and Girdler85,Reference Roca, Schmidt and Rubinow87,Reference Śliwerski, Koszałkowska, Mrowicka and Szafran89,Reference Forrester-Knauss, Zemp Stutz, Weiss and Tschudin95–Reference Halbreich and Endicott97,Reference Choi, Baek, Noh, Kim, Choi and Ha101,Reference El Dahr, de Azevedo Cardoso, Syan, Caropreso, Minuzzi and Smith102,Reference Slyepchenko, Frey, Lafer, Nierenberg, Sachs and Dias104,Reference Liang, Yang, Liao, Yang, Lin and Wu105,Reference Syan, Minuzzi, Smith, Costescu, Allega and Hall107} These included earlier age at onset of mood disorders or PMDD/PMS, more frequent and severe mood symptoms, higher rates of rapid cycling, psychiatric hospitalisations and suicidality. Notably, comorbidity was also linked to increased cognitive and functional impairment, regardless of diagnostic order. While most studies did not explicitly assess the temporal sequence of diagnoses over an extended period of time, those that did so suggested that PMDD/PMS may precede and predict the onset or worsening of mood disorders. ^{Reference Li, Tsai, Bai, Su, Chen and Liang75,Reference Hartlage, Arduino and Gehlert82,Reference Roca, Schmidt and Rubinow87} Collectively, these findings underscore the fact that comorbidity is not merely additive but may represent a distinct and more severe illness trajectory requiring targeted clinical attention.