Clinicians and patients suffering from major depression are confronted with the gap between guidelines produced by so‐called evidence‐based medicine and prescription patterns emerging from experience‐based medicine, as well as with the gap between artisanal prescribing and the siren song of personalization, stratification and precision medicine.
Perlis’ elegant paper describes the many challenges in abandoning personalization to get to precision in the pharmacotherapy of depression1. Currently, physicians indeed practice some form of personalization by taking into account the patient's symptom profile as well as the safety and tolerability of the different antidepressants, although taking the symptom profile into account is not or poorly empirically supported. Actually, the choice of a specific antidepressant is mainly based on the presence of a specific symptom (52%) or the wish to avoid a specific side effect (49%), and the specific symptoms considered are mainly anxiety (20%), insomnia (18%) and fatigue (14%)2.
Before antidepressant treatment can start moving from artisanal prescribing to precision medicine, several issues should be addressed. Taking more into account the “anima” (the individual, the real person, as well as the illness) and not only the “persona” (mask, character imposed by our diagnostic and assessment tools) seems to be mandatory.
That randomized clinical trials (RCTs) really represent the gold standard is questionable: “never before have the inadequacies of RCTs been so apparent to so many; yet, equally, never before have those in position of authority – from regulators, to policy‐makers, to doctors – relied so extensively on RCTs’ evidence”3. Efficacy estimates are usually based upon RCTs, but only about 10 to 20% of daily practice patients “fit into” exclusion and inclusion criteria.
Furthermore, efficacy estimates taken from RCTs heavily depend on study design: response rates of 52% and 34% for antidepressant and placebo, respectively, in two‐arm studies, 58% and 45% in three‐arm studies (two antidepressant arms, one placebo arm) and 65% in studies comparing two antidepressants; these differences can only be explained by the changing probability of receiving placebo: 50%, 33% and 0%, respectively4.
The role of patients’ expectations was also shown by a trial comparing sertraline, hypericum and placebo, which found no effect of assigned treatment on clinical improvement, but a significant effect of patient's guess on which treatment he/she was assigned to: patients who guessed taking sertraline or hypericum had significantly higher response rates (56% and 68%, respectively) than patients who guessed taking placebo (24%)5.
Many patients have ambivalent attitudes towards antidepressants that significantly influence outcome: patients with a rather negative, neutral or rather positive attitude towards taking antidepressants at baseline were found to have placebo response rates of 34%, 36% and 56%, respectively, and antidepressant response rates of 51%, 56% and 69%, respectively6.
Socio‐demographic characteristics are seldom taken into account, but variables such as living with other persons (versus living alone) or being unemployed (versus employed) dramatically influence the outcome of antidepressant treatment (OR: 2.81 and 0.27, respectively)7. There is also an ongoing debate on whether taking into account the patient's preference for pharmacotherapy or psychotherapy influences outcome. All these aspects should be considered before we try to improve our treatments for depression, be it by personalization, stratification or precision medicine.
In addition, the “persona” of the diagnostic criteria and of the assessment tools only partially represents the “anima” of the patient and of the depressive illness. A major depressive episode cannot be fully understood either by nine DSM or ten ICD criteria, or by ten Montgomery‐Åsberg Depression Rating Scale (MADRS), seventeen Hamilton Depression Rating Scale (HAMD) or thirty Inventory of Depressive Symptomatology (IDS) items.
One important limitation of the DSM criteria for major depressive episode is the massive heterogeneity they cause: almost endless combinations of criteria are possible. Indeed, when you need five out of nine criteria and, moreover, most of these criteria are compound (e.g., psychomotor retardation or agitation), two patients with major depressive episode can have no symptom in common. This of course hampers “personalized” treatment as well as clinical and etiopathogenetic research.
When assessing change during treatment, the standard rating scales face the same problems. Moreover, the HAMD covers a lot of associated anxiety and neurovegetative symptoms, while the IDS has a 16‐item version closely reflecting the DSM criteria and a 30‐item version adding commonly associated symptoms (anxiety, irritability) and items relevant to depression “subtypes”. DSM depression symptoms (included in the IDS 16‐item version) do not seem to be of higher clinical relevance than non‐DSM symptoms (additionally represented in the IDS 30‐item version) with respect to either their centrality (connectedness of each symptom with all other symptoms) or their relation to psychosocial functioning or life stress8.
Furthermore, the criteria/signs/symptoms we use for diagnosis and assessment do not reflect the patient's concerns. Who is the judge? It has been documented that physicians differ significantly from patients in what they consider important for “being cured for depression”. For physicians, the top five items are negative feelings, feeling down, little interest or pleasure, disrupted social life, and feeling tired, while for patients the top five items are “to what extent is life meaningful”, “how much do you enjoy life”, “how satisfied you are with yourself”, “how able you are to concentrate”, and “negative feelings”9. Patients do attach more importance to restoration of positive mood and cognitive functioning than to decrease of negative mood. However, standard rating scales do not assess positive mood.
We feel that, before we can move from artisanal prescription patterns into precision medicine, patients’ characteristics, beliefs and attitudes should be better taken into account, and diagnostic and assessment tools should be revised.
Koen Demyttenaere University Psychiatric Center, KU Leuven, Leuven, Belgium
References
- 1. Perlis RH. World Psychiatry 2016;15:228‐35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Zimmerman M, Posternak M, Friedman M et al. Am J Psychiatry 2014;161:1285‐9. [DOI] [PubMed] [Google Scholar]
- 3. McGoey L. Hist Hum Sci 2010;23:58‐78. [DOI] [PubMed] [Google Scholar]
- 4. Sinyor M, Levitt AJ, Cheung AH et al. J Clin Psychiatry 2010;71:270‐9. [DOI] [PubMed] [Google Scholar]
- 5. Chen JA, Papakostas GI, Youn SJ et al. J Clin Psychiatry 2011;72:1669‐76. [DOI] [PubMed] [Google Scholar]
- 6. Demyttenaere K, Reines EH, Lönn SL et al. Prim Care Companion CNS Disord 2011;13(4). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Demyttenaere K, Verhaeghen A, Dantchev N et al. Prim Care Companion J Clin Psychiatry 2009;11:307‐31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Fried EI, Epskamp S, Nesse RM et al. J Affect Disord 2016;189:314‐20. [DOI] [PubMed] [Google Scholar]
- 9. Demyttenaere K, Donneau AF, Albert A et al. J Affect Disord 2015;174:390‐6. [DOI] [PubMed] [Google Scholar]