In the well-articulated paper by Malhi et al.¹ in this journal, several problems with diagnosing bipolar disorder in children are discussed and as rightly pointed out “impede our ability to conduct meaningful research and advance clinical practice”. In fact, one could argue that the diagnostic challenges outlined apply to the diagnoses of mood disorders more generally. That is, reliance on a diagnostic checklist that reflects largely non-specific symptoms that cross diagnostic boundaries and are open to interpretation yield a highly heterogeneous population of mood disordered patients that share the same diagnosis but little else—differing in clinical course, family history, prognosis, treatment response and genetic and neurobiological correlates.

Malhi et al. offer a novel solution to the current diagnostic dilemma. The authors express hope that revising the current taxonomy so as to focus on developmentally sensitive symptom clusters, reflecting the evolution of the disorder over development, will advance the field past the current stalemate and improve diagnostic accuracy. While we agree that a developmental lens provides an informative perspective through which to view psychopathology, there is no evidence that a sole focus on symptoms, no matter how well developmentally nuanced, will improve diagnostic classification. Rather, substantive evidence supports the need to identify more homogenous subtypes from within the current heterogeneous bipolar diagnostic construct to advance risk prediction, pharmacotherapy, and discovery research. Three bipolar subtypes based on distinct clinical profiles have been described based on research extending over six decades, each with preferential response to stabilizing treatment with lithium, antipsychotics and antiepileptics, respectively (Figure 1).² Therefore, an alternative evidence-based solution would be to include these bipolar subtypes in a revised taxonomy.

Specifically, substantive evidence supports that a long-term response to lithium identifies a more homogeneous subtype of bipolar disorder characterized by a recurrent episodic course, complete remission, a history of episodic mood disorders in family members, and distinctive genetic correlates.³ This distinctive clinical profile, identified by multivariate analyses, was actually delineated by Kraepelin over a century ago. Further, prospective longitudinal studies of the offspring of lithium responsive (LiR) and lithium non-responsive (LiNR) bipolar parents have provided evidence that bipolar disorder debuts as a depressive episode in adolescence, years on average before emergence of the first manic episode.⁴ The developmental history and clinical course differ between subgroups, with offspring of LiRs having normal or gifted development and offspring of LiNRs manifesting neurodevelopmental disorders (ADHD, learning difficulties). Childhood clinical antecedents predicting major mood disorders include anxiety and sleep disorders, which in the offspring of LiRs follow an episodic course, while in offspring of LiNRs are chronic or fluctuating with incomplete remission (Figure 2).⁴ Further, the clinical course of mood disorders in offspring aligns with that of the parent; that is offspring of LiRs manifest episodic remitting mood disorders with stable functioning between episodes, while offspring of LiNRs manifest chronic or partial remitting mood disorders with lower global functioning over time.⁴ Self-reported manic symptoms did not differentiate high-risk from control offspring (of well parents)—in fact, controls endorsed higher hypomanic symptom levels; however, hypomanic symptoms identified on clinical assessment did predict onset of mood disorders in high-risk offspring, while no clinically meaningful hypomania was identified in controls.⁵

Taken together, longitudinal studies over decades of carefully prospectively studied adult bipolar patients and their relatives, including their children, have provided convergent evidence supporting bipolar subtypes that differ in characteristic developmental trajectories of emergent psychopathology, clinical course, prognosis, treatment response, and genetic and neurobiological correlates. This strongly suggests these clinical profiles index bipolar subtypes with shared genetic factors and pathophysiological mechanisms that differ meaningfully between subtypes. Therefore, we argue that a necessary revision to advance precision diagnosis that maps to preferential stabilizing treatment and reliably associated biomarkers rests on the incorporation of characteristic clinical profiles of bipolar subtypes into the diagnostic taxonomy. The clinical profiles, as illustrated briefly here, go beyond developmentally sensitive symptom clusters, which only have clinical meaning when considered in the context of a carefully detailed family history and clinical course.

Had the longitudinal evidence identifying characteristic clinical profiles of bipolar subtypes (including the developmental trajectories) been applied and incorporated in routine clinical practice, the entire debate about the validity of a pediatric (i.e. pre-pubertal mania) bipolar disorder equivalent might have been avoided or at least put to rest much earlier on. That said, lessons learned from the pediatric bipolar debate include that symptoms alone are insufficient evidence on which to rest a stable and accurate diagnosis, especially early in the emergent course. Further, as in other areas of medicine, the importance of a thorough clinical assessment that considers all predictive clinical information in the diagnostic formulation and includes collateral history and a carefully collected family history is paramount. While structured and semi-structured interviews and symptom checklists (developmentally sensitive or not) may be useful in large epidemiological studies, advances in psychiatry clinical practice and discovery research will require selective focus on carefully clinically characterized patients in order to identify those of the same bipolar subtype associated with a predictable course, preferential response to stabilizing treatment and shared pathogenesis.

在 Malhi 等人发表的清晰的论文中。^{在该杂志的第 1}期中，讨论了诊断儿童双相情感障碍的几个问题，并正确指出“阻碍了我们进行有意义的研究和推进临床实践的能力”。事实上，有人可能会说，所概述的诊断挑战更普遍地适用于情绪障碍的诊断。也就是说，对诊断检查表的依赖主要反映了跨越诊断界限并且可以解释的非特异性症状，产生了高度异质的心境障碍患者群体，他们具有相同的诊断，但除此之外几乎没有什么不同——临床病程、家族史、预后、治疗反应以及遗传和神经生物学相关性。

马尔希等人。为当前的诊断困境提供了一种新颖的解决方案。作者表示，希望修订当前的分类法，以关注发育敏感的症状群，反映疾病随发育的演变，从而推动该领域突破目前的僵局，并提高诊断准确性。虽然我们同意发展的视角提供了观察精神病理学的信息丰富的视角，但没有证据表明，无论发展的细微差别如何，仅关注症状会改善诊断分类。相反，实质性证据支持需要从当前异质双相诊断结构中识别更多同质亚型，以推进风险预测、药物治疗和发现研究。超过 60 年的研究已经描述了基于不同临床特征的三种双相亚型，每种亚型分别对锂、抗精神病药和抗癫痫药的稳定治疗有优先反应（图 1）。²因此，另一种基于证据的解决方案是将这些躁郁症亚型纳入修订后的分类法中。

具体来说，实质性证据支持，对锂的长期反应可识别出一种更同质的双相情感障碍亚型，其特征是反复发作的病程、完全缓解、家庭成员有发作性情绪障碍史以及独特的遗传相关性。³这种通过多变量分析确定的独特临床特征实际上是由 Kraepelin 在一个多世纪前描述的。此外，对锂反应性（LiR）和锂非反应性（LiNR）双相父母的后代进行的前瞻性纵向研究提供了证据，表明双相情感障碍在青春期以抑郁发作的形式首次出现，平均比首次躁狂发作出现数年。⁴亚组之间的发育史和临床过程有所不同，LiR 的后代具有正常或天才发育，而 LiNR 的后代则表现出神经发育障碍（ADHD、学习困难）。预测主要情绪障碍的儿童临床前因包括焦虑和睡眠障碍，在 LiR 的后代中，这些障碍呈间歇性病程，而在 LiNR 的后代中，这些障碍是慢性的或波动性的，且不完全缓解（图 2）。⁴此外，后代情绪障碍的临床病程与父母的一致；也就是说，LiR 的后代表现出阵发性缓解型情绪障碍，在发作之间具有稳定的功能，而 LiNR 的后代则表现出慢性或部分缓解型情绪障碍，随着时间的推移，整体功能较低。⁴自我报告的躁狂症状并不能将高风险儿童与对照后代（健康父母的）区分开来——事实上，对照后代认可较高的轻躁狂症状水平；然而，临床评估中发现的轻躁狂症状确实可以预测高风险后代情绪障碍的发作，而在对照中没有发现有临床意义的轻躁狂症状。⁵

总而言之，对成年双相情感障碍患者及其亲属（包括他们的孩子）进行了数十年的纵向研究，提供了支持双相情感亚型的一致证据，这些亚型在突发精神病理学、临​​床病程、预后、治疗反应以及遗传和遗传的特征发展轨迹上有所不同。神经生物学相关性。这强烈表明这些临床特征表明双相亚型具有共同的遗传因素和病理生理机制，而亚型之间存在显着差异。因此，我们认为，对推进精准诊断进行必要的修订，映射到优先稳定治疗和可靠相关的生物标志物取决于将双相亚型的特征性临床特征纳入诊断分类法。正如此处简要说明的，临床特征超出了发育敏感的症状群，只有在仔细详细的家族史和临床病程的背景下考虑时，这些症状群才具有临床意义。

如果识别双相亚型特征性临床特征（包括发育轨迹）的纵向证据被应用并纳入常规临床实践，那么关于儿科（即青春期前躁狂）双相情感障碍等效物有效性的整个争论可能就可以避免或至少更早地休息。也就是说，从儿科双相情感障碍辩论中吸取的教训包括，仅凭症状不足以得出稳定而准确的诊断，尤其是在紧急病程的早期。此外，与其他医学领域一样，考虑诊断方案中所有预测性临床信息并包括附带病史和仔细收集的家族史的彻底临床评估至关重要。虽然结构化和半结构化访谈和症状检查表（发育敏感或不敏感）可能在大型流行病学研究中有用，但精神病学临床实践和发现研究的进展将需要选择性地关注仔细临床特征的患者，以便识别那些患有相同双相情感障碍的患者亚型与可预测的病程、对稳定治疗的优先反应和共同的发病机制相关。