Overview
My research questions the boundaries between mental disorders by identifying symptoms that cut across depression, anxiety, and personality disorders, yet also seeks a solution to improve diagnosis and ultimately lead to more effective treatments. I study how diagnostic constructs are formulated and validated, and how they are used in research and in clinical practice. I am working to help move the field from using diagnostic syndromes constructed from clinical description to one that incorporates an empirical understanding of core psychological and biological mechanisms. My work includes studying the longitudinal course of psychopathology, studying how clinicians conceptualize and diagnose mental disorders, developing a framework to guide psychopathology research, and clarifying the structure and mechanisms of psychopathology by researching trans-diagnostic constructs. Perhaps the most important impact of my work on the field has come from my mentoring of undergraduate students (many of whom are now in first-rate graduate programs, as well as those who have completed their graduate training and are working in the field as clinicians and clinician scientists); my post-doctoral fellows (many of whom now hold faculty positions); and early career faculty whom I have mentored (many of whom have been promoted). I believe that imparting new ways of thinking about how to research and treat mental illness in the context of historical efforts offers the best chance to prepare future clinician-scientists to better define psychopathology constructs, and to develop new approaches to assessment and diagnosis that will speed the development of better treatments and ultimately better relieve human suffering from mental illness.
Research Trajectory as a Clinical Scientist
Since it began in 1996 and for the entirety of its project period, I was a member of the research team of the NIMH-funded Collaborative Longitudinal Study of Personality (CLPS). The CLPS project was a prospective, naturalistic study of a clinical sample selected with a principal diagnosis of one of four personality disorders (Borderline, Schizotypal, Avoidant, and Obsessive-Compulsive) or major depression. The scale of this study required the collaboration of group clinical scientists working together across multiple universities (Brown, Columbia, Harvard, Texas A&M, and Yale). We recruited a large sample of patients (N = 668) and followed them over a ten-year period, collecting extensive clinical data on each participant. The implications of the findings have been far-reaching and raised questions about the validity of DSM-defined personality disorders and about the relationship of personality disorders to other mental disorders as well as the boundaries among them. Findings also demonstrated the public health implications of impaired functioning, increased treatment utilization, and suicidal behavior in patients with severe personality pathology. (For an overview of CLPS, see Skodol, Gunderson, Shea, McGlashan, Morey, Sanislow, et al., 2005.) We demonstrated that remission from DSM-IV defined personality disorders was more common than postulated by DSM-definitions (Grilo, Sanislow, et al., 2004; Gunderson, Bender, Sanislow, et al., 2003). We hypothesized that personality disorders could be conceptualized as a “hybrid” model of personality traits and psychiatric symptoms (McGlashan, Grilo, Sanislow et al., 2005). Our study raised pivotal questions about the way that personality disorder constructs are defined (Sanislow et al., 2002a; Sanislow et al., 2009). For example, when I examined ten years of prospective diagnostic data, the results disputed the common assumption that the structure and boundaries of personality disorders are chronic and stable (Sanislow et al., 2009). Instead, manifestations of DSM personality constructs in patients lose their structure over time as the criteria become more correlated across disorders.
In my work, I have also examined how people think about mental disorders. Clinicians and laypeople alike tend to see mental disorders as less real than physical disorders, and less a result of natural causes (Ahn, Flanagan, Marsh & Sanislow, 2006). Perhaps this drives a need to embrace DSM categories because such constructs may seem tangible. In the CLPS project we demonstrated that dimensional, trait-based models of personality disorders (an alternative to the DSM) are more stable and valid than the DSM personality disorder definitions (Morey, Warner, Shea, Gunderson, Sanislow, et al., 2003), and we have shown that DSM personality disorder diagnoses made by clinicians have less predictive validity compared to structured interview or self-report trait models (Samuel, Sanislow, et al., 2013). Yet in other research, when we asked clinicians to make diagnoses with dimensional trait-based models, they encountered more difficulty and reported less satisfaction when using these alternative models compared to using the DSM (Rottman, Ahn, Sanislow, & Kim, 2009; Rottman, Kim, Ahn, & Sanislow, 2011). It may be that mere familiarity with the DSM is why clinicians are more comfortable using it. In any event, clinicians are disinclined to use these dimensional models even though evidence suggests that dimensional models offer a better approach. Our work is provocative, and raises important questions for clinical practice.
Results from my research have led me to be concerned with problems of contemporary diagnosis arising from expert consensus and based on clinical syndromes (see Sanislow et al., 2010; Sanislow, Quinn, & Sypher, 2015; Sanislow, Morris, Pacheco, & Cuthbert, 2020). In clinical settings, the reality is that most patients meet criteria for several diagnoses, and it is not clear that these are separate labels are actually separate entities. I’ve studied how wide or narrow a diagnostic boundary should be (Sanislow et al., 2000), and investigated comorbidity (e.g., Grilo, Sanislow, & McGlashan, 2002) and the specificity and coverage of DSM diagnoses (e.g., Grilo, Sanislow, et al. 2007). One way of clarifying such issues is to examine the diagnostic boundaries between avoidant personality disorder and social phobia (Ralevski, Sanislow, et al., 2009; see also Sanislow, Bartolini, & Zoloth, 2012; Sanislow, da Cruz, Gianoli, & Reagan, 2012; Sanislow & Hector, 2020a & b). I have also addressed heterogeneity within diagnoses. For instance, two people can be diagnosed with borderline personality disorder but share only one symptom in common. I have identified intermediate phenotypes (a.k.a. “endophenotypes”) that represent core components of the descriptive diagnosis by developing a three-factor model that posits a dynamic relationship between affective dysregulation, behavioral dysregulation, and disturbed relatedness (Sanislow et al., 2000; Sanislow et al., 2002b). This model has received much interest and generated empirical studies from other labs because of the promise that cross-cutting dimensions such as these (also called intermediate phenotypes) may be more directly connected to specific mechanisms than the DSM syndrome as a whole.
In my work, neuroimaging (MRI, PET, EEG) and behavioral approaches complement clinically based diagnostic studies. These approaches, aimed at identifying putative mechanisms, provide clues to help elucidate core components to help clarify boundaries of clinical disorders, in addition to raising new questions. I have used these approaches to contrast comorbid disorders. A seminal fMRI study found laterality differences in emotion processing for patients diagnosed with borderline and PTSD, compared to those without (Donegan, Sanislow, et al., 2003). Furthermore, patients diagnosed with borderline and with PTSD showed patterns consistent with prior studies of PTSD but those patients diagnosed with borderline and not with PTSD were more emotionally reactive than the those diagnosed with PTSD. This raised the question of whether a diagnosis of borderline disorder adds anything to a PTSD diagnosis for these patients given the overlap in descriptive criteria for the two disorders.
More generally, my work provides evidence for cognitive components and associated neural processes relevant to psychopathology, for example, by using cognitive tasks to study disruptions in the cognitive processes. My premise is that identifying disruptions in attention, working memory, or long-term memory processes will yield clearer understanding of psychopathology and more specific targets for treatment than those based on clinical description. For this, I have studied how emotion affects cognitive processes in normal participants behaviorally and using fMRI (Johnson, Raye, Mitchell, Greene, Cunningham, & Sanislow, 2005, exp. 6A and 6B; Johnson, Mitchell, Raye, McGuire, & Sanislow, 2006). In my laboratory, I have also developed stimulus materials for studying emotion and for mood induction (e.g., Gabert-Quillen, Bartolini, Abravanel, & Sanislow, 2015). Work in my lab also uses event-related potential (ERP) techniques to compare neural responses to errors among individuals who, for instance, vary in rumination tendencies, with the goal of elucidating how and why rumination arises (Tanovic, Hajcak, & Sanislow, 2017), as well as the relation of rumination to depressive symptoms (Berry, Tanovic, Joormann, & Sanislow, 2019). We are also using ERP to study neural correlates of attention and error detection in decision making (Patalano, Loli, Sanislow, 2018; 2020).
The problems that I have identified in the study of personality and mental disorder definitions and diagnoses, and by researching psychopathology-related mechanisms using behavioral and neuroimaging approaches, have motivated my broad perspective on understanding psychopathology. On leave from Yale (2008-2009), my work as Chief for the NIMH Mood Disorders and Sleep Research Program (2008-2009) was a pivotal transition for my research program. At NIMH, I served in the translational division and served in roles that bridged the basic behavioral science division and the services and intervention division. I was called on to advise on matters that included identifying clinically meaningful biomarkers in basic research, and my work also included helping to define targets for clinical trials in service research. A view of clinical science at the national level provided me a large-scale understanding of the state of the field, and clarified my focus on problems with psychiatric diagnosis. It was evident that better integration of behavioral, psychological, and biological systems could make it possible to develop a nosology based on aberrations in these mechanisms rather than trying to identify aberrations based on our current nosology based on clinical syndromes.
Because of my expertise spanning cognitive neuroscience and clinical problems, and my research raising questions about the DSM disorders, I was selected as member of the NIMH Internal Working Group to develop new ways to classify mental illness for research purposes. This initiative became the Research Domain Criteria (RDoC), and is changing the paradigm for how mental illness is researched (e.g., Insel, Cuthbert, Garvey, Heinssen, Pine, Quinn, Sanislow, & Wang, 2010; Sanislow et al., 2010; Sanislow, Quinn, & Sypher, 2015; Sanislow, 2016a & Sanislow, 2016b; Sanislow, Morris, Pacheco, & Cuthbert, 2020), it has occupied a central focus of my work for over the most recent decade of my career (Sanislow, 2020). As a member of the RDoC Internal Working Group, I helped to develop the structure of RDoC. In six meetings over two years, we brought together leaders with expertise in each of the RDoC domains to critically review empirical support for putative mechanisms of psychopathology within the scope of each domain. Those meetings helped define constructs by evaluating the empirical evidence and the relevance for psychopathology. I co-authored the summaries of the workshops which are published on the NIMH website. My ongoing work on RDoC has focused on ways to illustrate to the field how to understand and implement RDoC-framed research, as well ways to improve the RDoC framework. For instance, I participated in a National Advisory Mental Health Council workgroup to evaluate the addition of a motor systems domain, as well as meetings that focused on the development of valid and reliable behavioral tasks for RDoC constructs.
The RDoC provides a framework to answer clinical research questions central to my work: the problem of high rates of comorbidity of mental disorders, the relation of clinical phenomena to basic cognitive and neural mechanisms, the observation that mental disorders involve multiple disruptions in putative mechanistic factors (e.g., disruptions in fear, impulse control, and cognition-emotion interactions), and the observation that some disrupted mechanisms appear to play major roles in many disorders (Sanislow et al., 2010; Sanislow, 2020). It is exemplified in transdiagnostic approaches, studying cross cutting features of psychopathology such as rumination as I do in my lab (e.g., Tanovic, Hajcak, & Sanislow, 2017) as well as more complex cognitive processes such as decision making (Patalano, Lolli, & Sanislow, 2020). Using dimensions of such features (instead of DSM disorders) as independent variables illustrates an implementation of the RDoC framework. As a continuing member of the NIMH RDoC Internal Working Group, I have a hand in moving the field toward a new way to understand mental disorders, while in my lab, I am able to work on pieces of the puzzle. As we approach the ten-year anniversary of RDoC, its influence is clear, as is the promise that it will inform future versions of clinical diagnostic systems (Sanislow, 2016a; Sanislow, 2020).
My work is motivated by decades of clinical practice, providing clinical assessment and intervention to those suffering the kinds of psychopathology that I research, and by teaching psychology interns, post-docs, and psychiatric residents to do the same. Early on, my research emphasized the importance of studying personality disorder treatment outcomes (Sanislow & McGlashan, 1998). Before that, the study of treatment outcomes was rare; now, it has become routine. My theoretical and empirical contributions have called into question the belief that personality disorders are immutable thus providing a direction for the development of treatments and providing hope for people who suffer personality pathology (Sanislow, Marcus, & Reagan, 2012). The realization that we need other ways to define psychopathology constructs came from studying personality and other mental disorders, and grappling with the problems of comorbidity and heterogeneity of syndromes defined by grouping clinically described symptoms (e.g., Sanislow et al., 2000; Sanislow et al., 2002a & b; Sanislow et al., 2009). To better integrate mechanisms—psychological, behavioral, or neural—we need a different starting point, one that allows the investigation of a fuller range of clinical phenomena as well as the variability in the related mechanism(s) of dysfunction. The best targets of study may not be the familiar, contemporary diagnoses but rather neural, cognitive, or behavioral mechanisms, and the best approach to capture palpable psychopathology need not be constrained by current nosology (Sanislow et al., 2010; Sanislow, 2016a & b; Sanislow, Morris, Pacheco, & Cuthbert, 2020).
