Personalized Medicine in Psychiatry: from DSM to brain-based RDoC, iSPOT‑D and biomarkers

(Edi­tor’s Note: this is Part 2 of the new 3‑part series writ­ten by Dr. Evian Gor­don draw­ing from his par­tic­i­pa­tion at the Per­son­al­ized Med­i­cine World Con­gress on Jan­u­ary, 23, 2012 at Stan­ford University.)

Most Per­son­al­ized Med­i­cine research in Psy­chi­a­try using mol­e­c­u­lar mea­sures alone have failed to repli­cate. Whilst dis­ap­point­ing, this is not sur­pris­ing, since 80% of human 25,000 genes have some effect on the brain.

There are there­fore grow­ing efforts expand­ing Genom­ic Bio­mark­ers in Psy­chi­a­try to Neu­roimag­ing (all Brain-based bio­log­i­cal and cog­ni­tive mea­sures). Some approach­es tar­get Gene+Brain Com­bi­na­tion Bio­mark­ers to spe­cif­ic neu­ro­trans­mit­ter “Cir­cuits. For exam­ple the Sero­tonin cir­cuit: since SSRIs impact upon this Amyg­dala-Ante­ri­or Cin­gu­late-Raphe Nucleus–BDNF mol­e­c­u­lar mod­u­la­tion circuit.

What­ev­er the approach, the goal is to shed light on bio­log­i­cal­ly rel­e­vant char­ac­ter­is­tics of the patient and then increas­ing­ly elu­ci­date which of these pre­dict a spe­cif­ic treat­ment response.

Diverse groups are shap­ing this bio­mark­er land­scape for psy­chi­a­try, includ­ing DSM‑5, NIMH and many trans­la­tion­al neu­ro­science groups.

I pre­sent­ed a dis­til­la­tion of these in my PMWC presentation:

• The need for neu­ro­bi­o­log­i­cal infor­ma­tion to “val­i­date” con­structs of Psy­chi­atric disorders;
• The need for mul­ti­ple com­ple­men­tary Gene+Brain mea­sures across scale;
• A Neu­rode­vel­op­men­tal con­text, with empha­sis on the age of onset and peak peri­ods when Psy­chi­a­try insta­bil­i­ties manifest;
• A Dimen­sion­al con­text of Psy­chi­atric Dis­or­ders, where under­ly­ing neu­ro­bi­ol­o­gy is con­tin­u­ous from nor­mal­i­ty to disorder;
• Large Data­bas­es of norms for age and lon­gi­tu­di­nal clin­i­cal out­comes with clear end-points that can confirm/disconfirm Bio­mark­er treat­ment predictions;
• A trans­la­tion­al approach that links basic mech­a­nism of Psy­chopathol­o­gy to tar­get­ed Bio­mark­ers that pre­dict who is most like­ly to respond best to what intervention.

These empha­sizes are reflect­ed in the move to the DSM‑5 and the “Research Domain Research Cri­te­ria (RDoC) ini­ti­atve launched by the Nation­al Insti­tute of Men­tal Health.

The goal of DSM‑5 to expand “signs and symp­toms” clas­si­fi­ca­tion to incor­po­rate bio­log­i­cal mea­sures, is cap­tured to some extent by this quote from Task Force Chair, David Kupfer:

“As we grad­u­al­ly build on our knowl­edge of men­tal dis­or­ders, we begin bridg­ing the gap between what lies behind us (pre­sumed eti­olo­gies based on phe­nom­e­nol­o­gy) and what we hope lies ahead (iden­ti­fi­able patho­phys­i­o­log­ic eti­olo­gies).” Amer­i­can J Psy­chi­a­try, 168 (7): 672–674, 2011.

The DSM pro­vides a con­sis­tent frame­work for clin­i­cal com­mu­ni­ca­tion. But it was devel­oped at a time of lim­it­ed knowl­edge in brain sci­ence. Its cur­rent lim­i­ta­tion in Per­son­al­ized treat­ment pre­dic­tion reflects this knowl­edge gap and the grow­ing call for flesh­ing out the clin­i­cal real­i­ty of co-mor­bid­i­ty and het­ero­gene­ity by con­nect­ing them with mean­ing­ful bio­log­i­cal entities.

RDoC is a com­ple­men­tary approach (Insel et al. 2010. Amer­i­can J Psy­chi­a­try, 1020; 167 (7): 748–751). One of its goals will be to inform new edi­tions of clas­si­fi­ca­tion sys­tems. But the breadth of its impli­ca­tions con­struc­tive­ly chal­lenges all cur­rent con­ven­tion­al approach­es in Psychiatry.

RDoC is a Frame­work of 5 orga­niz­ing ‘Domains’ that cut across tra­di­tion­al dis­or­der clas­si­fi­ca­tions. The Research Domain Cri­te­ria (RDoc) are:

Neg­a­tive Valence Sys­tems (Threat and Anxiety);
Pos­i­tive Valence Sys­tems (Rewards and Habits);
Cog­ni­tive Sys­tems (atten­tion, per­cep­tion, work­ing memory);
Sys­tems for Social Process­es (iden­ti­fi­ca­tion of facial expres­sion, The­o­ry of Mind);
Arousal/Regulatory Sys­tems (State and Trait dynamics).

By expli­cat­ing RDoC, NIMH seeks to elu­ci­date the under­pin­ning bio­log­i­cal mech­a­nisms across scale (from genes to cir­cuits and behav­ior) of these 5 domains. This is antic­i­pat­ed to lead to new mol­e­c­u­lar and neu­roimag­ing drug dis­cov­ery tar­gets, accel­er­ate a new cul­ture of research strate­gies (suc­ceed or fail fast; stan­dard­iza­tion, inte­gra­tion, data shar­ing) and pro­vide fun­da­men­tal­ly new bio-behav­ioral approach­es to clas­si­fy­ing (and treat­ing) men­tal disorders.

Using stan­dard­ized meth­ods to advance Per­son­al­ized Med­i­cine in Psychiatry

One approach to inte­grat­ing bio­mark­ers with clin­i­cal knowl­edge in Psy­chi­a­try is the use of stan­dard­ized mea­sure­ment to link each type of information.

Brain Resource has a flag­ship study under­way that uses stan­dard­ized meth­ods to iden­ti­fy bio­mark­ers of per­son­al­ized treat­ment for depres­sion: the inter­na­tion­al Study to Pre­dict Opti­mized Treat­ment for Depres­sion (iSPOT‑D).

iSPOT‑D uses the stan­dard bat­tery of mea­sures estab­lished by Brain Resource, and used to set up the inter­na­tion­al data­base. These mea­sures include:

Mol­e­c­u­lar (Genet­ic variants);
Brain Imag­ing (struc­tur­al and func­tion­al scans) ;
Phys­i­ol­o­gy (EEG, Evoked Poten­tials, heart rate and skin conductance);
Cog­ni­tive and Emo­tion­al behavior;
Self-report­ed experience.

iSPOT‑D is using these mea­sures to iden­ti­fy bio­mark­ers that pre­dict these end points:

• Over­all Response to the med­ica­tion (No/Yes: defined by symp­tom remission);
• Dif­fer­ent respons­es to the 3 most com­mon­ly employed anti­de­pres­sives in the U.S. (Esc­i­talo­pram; Ser­tra­line; Venlafaxine-XR);
• Dose (For Exam­ple Ven­lafax­ine: SSRI 150mg);
• Side effects;
• Long Term Remission.

These pro­to­cols have been pub­lished (For details, see: Williams, Rush, Koslow, Wis­niews­ki, Coop­er, Nemeroff, Schatzberg, Gor­don, Tri­als, 4, 2011).

iSPOT‑D will have 2,000 patients with Major Depres­sion Dis­or­der (MDD). The goal is to deter­mine the best bio­mark­ers for treat­ment pre­dic­tion in the 3 med­ica­tions. Data for the first 1,000 par­tic­i­pants is cur­rent­ly in analysis.

The need for bio­mark­ers in Psychiatry

The need for iden­ti­fy­ing bio­mark­ers is enor­mous, and depres­sion is a strik­ing exam­ple of this need.

The cost of depres­sion in the U.S. is $43.7 bil­lion per year, and $83 bil­lion when lost work pro­duc­tiv­i­ty is fac­tored in. Each year, over 60 mil­lion pre­scrip­tions are writ­ten for esc­i­talo­pram, ser­tra­line and ven­lafax­ine-XR. Retails sales for these med­ica­tions total $6 billion.

The exam­ple below serves to illus­trate the impli­ca­tions of dis­cov­er­ing Bio­mark­ers that pre­dict treat­ment response for the 3 most com­mon­ly used anti­de­pres­sants in the U.S.:

Let’s assume the cur­rent 1 in 3 chance of being pre­scribed the right anti­de­pres­sive treat­ment first time can be improved by 25% (from 33% to 42%). That means that around 2m peo­ple (10% of the 19m suf­fer­ing depres­sion in the U.S.) will ben­e­fit from sav­ing at least one incor­rect pre­scrip­tion. The iSPOT med­ica­tions rep­re­sent a 40% mar­ket share so that implies 800,000 peo­ple (assum­ing sim­ple aver­ages) will ben­e­fit from this sav­ing. Assum­ing an aver­age $50 pre­scrip­tion cost per patient, this implies a direct med­ica­tion cost sav­ing of $40m. This of course ignores the flow on costs of clin­i­cal con­sul­ta­tion time, pro­duc­tiv­i­ty loss­es and over­all patient suf­fer­ing when get­ting the wrong med­ica­tion. Adding the cost of even 1 lost day and 1 inef­fi­cient clin­i­cal vis­it mean and these sav­ings can increase ten fold (assum­ing $500 cost to an employ­er for one lost day and $100 for a clin­i­cal vis­it). It also ignores the ben­e­fit that a patient that reach­es a solu­tion faster will have high­er com­pli­ance tak­ing the med­ica­tion and data has shown that this leads to over­all reduced med­ical expen­di­tures of more than $1,000 pa (“Recov­ery from Depres­sion, Work Pro­duc­tiv­i­ty, and Health Care Costs Among Pri­ma­ry Care Patients, Gre­go­ry E. Simon, MD, et al. Gen­er­al Hos­pi­tal Psy­chi­a­try 22, 153–162, 2000).

Find­ing the right bio­mark­ers for each per­son is an impor­tant way to get the right med­ica­tion for each patient quickly.

To Be Continued…

New Series on Per­son­al­ized Med­i­cine and the Brain:

• Mon­day, Feb­ru­ary 13th: The State of Per­son­al­ized Medicine
• Mon­day, Feb­ru­ary 20th: Per­son­al­ized Med­i­cine in Psy­chi­a­try (above)
• Mon­day, Feb­ru­ary 27th: Work­ing with Health Care Indus­try Stake­hold­ers Towards Brain-based Per­son­al­ized Medicine

– Dr Evian Gor­don is the Exec­u­tive Chair­man of theBrain Resource Com­pany.  He ini­tially drew upon  his sci­ence and med­ical back­ground to estab­lish the inter­dis­ci­pli­nary Brain Dynam­ics Cen­ter, in 1986.  Through the Brain Dynam­ics Cen­ter and its col­lab­o­ra­tive net­works, Dr Gor­don estab­lished an “inte­gra­tive neu­ro­science” approach, ground­ed in the use of stan­dard­ized meth­ods across mul­ti­ple types of data. Using this approach, Dr Gor­don found­ed the “Brain Resource Com­pany”, that cre­ated the first inter­na­tional data­base on the human brain. The data­base is the asset which under­pins the devel­op­ment of new tools for brain health and its per­son­al­ized appli­ca­tion in the mar­ket, such as assess­ments of brain health, deci­sion sup­port sys­tems, and per­son­al­ized train­ing pro­grams. Brain Resource has also sup­ported the for­ma­tion of a non-prof­it 501c3 Foun­da­tion, called ‘BRAIN­net” (www.BRAINnet.net), through which sci­en­tists have access to many of these datasets for inde­pen­dent research.