BRITISH JOURNAL OF PSYCHIATRY (2006),(2006),188,180^185 188, 180^185

AUTHOR’ S P ROOF

Brain white-matter hyperintensities and treatment Participants were required to have a score of 16 or over on the 17-item Hamilton Rating outcome in major depressive disorder Scale for Depression (HRSD; Hamilton, 1967) at the screening visit. We excluded people who failed to respond during the DAN V. IOSIFESCU, PERRY F. RENSHAW, IN KYOON LYOO, HO KYU LEE, current episode of depression to at least ROY H. PERLIS, GEORGE I. PAPAKOSTAS, ANDERW A. NIERENBERG one adequate antidepressant trial. and MAURIZIO FAVA Treatment-resistant major depressive disorder Nineteen participants (mean age 39.6 years, s.d.s.d.¼9.8), of whom 10 (52.6%) were womenwomen,, were recruited at the Depression Background An increased incidence The association between major depressive Clinical and Research Program at Massa- of brain white-matter hyperintensities has disorder and increased prevalence of brain chusetts General Hospital for a clinical trial white-matter hyperintensities has been re- in people with treatment-resistant depres- been describedinmajordepressive ported in elderly people (Krishnan et aletal,, sion, beginning with open-label nortripty- disorder, butthebuttheimpactof impactof such 1997; de Groot et aletal, 2000), but studies in line (Nierenberg et aletal, 2003). Patients hyperintensities on treatment outcome is younger patients have been inconclusive eligible for inclusion were men and women still controversial. (Coffey(Coffey et aletal, 1993; Lenze et aletal, 1999; Lyoo aged 18–70 years with major depressive et aletal, 2002). In samples of elderly people disorder diagnosed using the SCID–P and Aims ToinvestigateTo investigate the relationship of with depression, such hyperintensities were a score on the HRSD of at least 18. brain white-matter hyperintensities with associated with lower rates of response to Treatment resistance was defined as non- cardiovascular risk factors and with antidepressant treatment (Hickie et aletal,, response to at least one, but no more than 1997; Simpson et aletal, 1998; Steffens et aletal,, five, adequate antidepressant trials during treatmenttreatmentoutcome outcome ininyounger younger people 2002) as well as higher rates of relapse dur- the current depressive episode. with major depressive disorder. ing follow-up (O’Brien et aletal, 1998; Yanai etet alal, 1998). The study reported here is, to our Exclusion criteria MethodMethod We assessed brain white- knowledge, the first to explore the impact For both the depressive disorder samples, matter hyperintensities and cardiovascular of white-matter hyperintensities on anti- exclusion criteria were , risk factorsin 84 people with major depressant treatment outcome and the psychotic disorder, a history of organic depressive disorder prior to initiating relationship between such hyperintensities mental or seizure disorder, serious or un- and cardiovascular risk factors in younger stable medical illness, substance misuse or antidepressanttreatment.We also people with depression. We predicted that dependence disorder active within the past assessed hyperintensitiesin 35 matched the presence of brain white-matter hyper- 12 months, acute suicidal risk, pregnancy, controls. intensities would correlate with cardiovas- lactation, history of adverse reaction or al- cular risk factors and with lower rates of lergy to the study medications, concomitant ResultsResults We found no significant treatment response. use of psychotropic medications, clinical or difference in the prevalence of white- laboratory evidence of thyroid abnormal- matter hyperintensities betweenbetweenthe the METHOD ities, an existing diagnosis of dementia or a score below 27 on the Mini-Mental State depression and the control groups.Left- Participants Examination (Folstein et aletal, 1975), and any hemisphere subcorticalsubcorticalhyperintensities hyperintensities All study participants signed consent forms, contraindication to magnetic resonance correlated withlowerwith lower rates oftreatmentof treatment approved by the institutional review board, imaging, including metallic implants or response.We found no correlation prior to the initial study visit. severe claustrophobia. between globalhyperintensitymeasuresglobalhyperintensity measures Non-treatment-resistant major depressive Controls and clinical outcome.Brain white-matter disorder We also recruited through advertisements hyperintensities correlated with The study group with non-treatment- 35 healthy volunteers, matched for age hypertension and age and withtotal resistant major depressive disorder com- and gender (40% females; mean age 39.3 cardiovascular risk score. prised 65 people aged 18–65 years (mean years, s.d.years,s.d.¼9.8). These volunteers under- age 40.7, s.d.¼10.2), recruited through went the physician-administered SCID–P Conclusions SubcorticalSubcorticalwhite-matter white-matter advertisements and clinical referrals for a to rule out any Axis I psychopathology. hyperintensities in the left hemisphere (but clinical trial beginning with open-label fluoxetine treatment (Fava et aletal, 2002).,2002). Tests and procedures notin other brain areas) may be associated All participants, 24 (37%) of whom were We assessed cardiovascular risk factors in with poor response to antidepressant women, met criteria for major depressive all participants with major depressive dis- treatmentin major depression. disorder, diagnosed by the physician- order following the National Institutes of administered Structured Clinical Interview Health Adult Treatment Panel III guide- Declaration of interest None. for DSM–III–R Axis I Disorders – Patient lines, based on the Framingham Heart Funding detailed in Acknowledgements. edition (SCID–P; Spitzer et aletal, 1989).,1989). Study (Wilson et aletal, 1998; Expert Panel

180180

Downloaded from https://www.cambridge.org/core. 28 Sep 2021 at 17:13:13, subject to the Cambridge Core terms of use. HYPERINTENSITIES AND TREATMENT OUTCOME IN DEPRESSION

AUTHOR’SAUTHOR’ S PROOFP ROOF

on Detection, Evaluation and Treatment of and fluid-attenuated inversion recovery The differences in the severity of white- High Blood Cholesterol in Adults, 2001). (FLAIR) axial images (TE¼133 ms,133ms, matter hyperintensities between partici- We recorded, for each person in the two TRTR¼9002 ms, inversion time TI¼2200 ms,2200ms, pants in the two depression groups and depression groups, age, gender, smoking 25625666192 matrix; field of view 22 cm, the healthy comparison sample were ana- status, family history of cardiovascular 5 mm thick slices, 2 mm skip). Voxel dimen- lysed usinglysedusing ww22-tests. We used multiple ordi- disease, total serum cholesterol,personal sions were 0.975 mm660.975 mm0.975mm663.0 mm.3.0mm. nal logistic regression to test the association history of arterial hypertension and diabetes,diabetes, Analysis of the images was performed off- between hyperintensity scores and cardio- and concomitant medications. We calcu- line using a SUN Microsystems (Mountain- vascular risk factors. Analysis of variance lated a cumulated cardiovascular risk score view, California, USA) Sparc2 workstation was used to test the association between (range 0–6) by assigning a point for each of and the radiological film. Lesions were clas- hyperintensity scores and the total cardio- the following six factors: sified according to the Fazekas classification vascular risk score (sum of the six (a)(a)age:age: male 45 years old or more, female system (Fazekas et aletal, 1987), which pro- cardiovascular risk factors). Since age is 55 years old or more; vides an assessment of severity of the one of the cardiovascular risk factors, these white-matter hyperintensities, rated sepa- analyses were not adjusted for age. Correla- (b)(b)activeactive smoking status within the past 12 rately for the subcortical white matter tions between clinical outcome variables months;months; (range 0–3) and the periventricular white (response and remission) and hyper- (c)(c)familyfamily history of premature vascular matter (range 0–3). The total white-matter intensity scores were tested using logistic disease (e.g. myocardial infarction, hyperintensity score was considered to be regression, adjusted for age. Statistical stroke) in male first-degree relatives the higher of the subcortical score and the significance was defined as PP550.05,0.05, under 55 years old, and in female periventricular score, following previous two-tailed.two-tailed. first-degree relatives under 65 years old; classifications of white-matter hyperinten- (d)(d)totaltotal cholesterol level greater than sity in people with major depressive dis- 5 mmol/l5mmol/l;; order (Krishnan et aletal, 1997). All ratings of RESULTSRESULTS (e)(e)arterialarterial hypertension (blood pressure white-matter hyperintensities were done by over 140/90 mmHg, or use of anti- an experienced neuroradiologist (H.K.L.), The demographic and clinical characteris- hypertensive medication); who was unaware of participant identity tics of the three study groups are presented in Table 1. The initial HRSD scores for (f)(f)diabetesdiabetes mellitus. and clinical status. Another investigator in the study (D.V.I.) assessed independently a depression severity were not statistically After the initial evaluation the non- selection of 111 magnetic resonance images different between the two depression treatment-resistant sample entered 8 weeks using the same rating criteria, for measure- groups: 21.6 v.v. 20.2; unpaired tt-test,-test, of open-label treatment with fluoxetine ment of interrater reliability. This was very PP¼0.15.0.15. 20 mg daily, following a 1-week wash-out good: the number of observed agreements The incidence of total brain white- phase. The HRSD was administered at each was 98 (88.3%) and weighted kk¼0.82.0.82. matter hyperintensities in the group with study visit (screen, baseline and then every The presence of severe hyperintensities non-treatment-resistant depressive disorder other week for 8 weeks). The treatment- was defined as a Fazekas scale score of 2 (63%) was not statistically different from resistant group were prescribed nortripty- or over, whereas scores below 2 were cate- that of the treatment-resistant group line at an initial dosage of 25 mg, which gorised as not severe, following previous (53%) or the control group (60%); was increased by 25 mg per day until a 22 classifications in people with major ww ¼0.881,0.881, PP¼0.83. Also, the incidence of dosage of 100 mg was reached, unless depressive disorder (Krishnan et aletal, 1997).,1997). severe brain hyperintensities in the non- patients were unable to tolerate the dosage In addition to the Fazekas scale scores, we treatment-resistant group (8%) was not increase because of side-effects. Blood determined the localisation of hyper- statistically different from that of the levels of nortriptyline were measured at intensities by hemisphere (left or right). treatment-resistant group (5%) or the con- weeks 2 and 6, and dosage adjustments 22 Subcortical hyperintensities were also trol group (6%); ww ¼0.364,0.364, PP¼0.95. As ex- were made after the second week if blood localised as being in the frontal lobe or pected, the individual’s age correlated with levels were below 400 nmol/l. Participants not in the frontal lobe area, using the the severity of brain hyperintensities then maintained their dosage of nortripty- central sulcus as a boundary. (ANOVA, d.f.¼117,117, FF¼11.0,11.0, PP¼0.0012).0.0012). line for 6 weeks. The HRSD was adminis- In all three study groups the majority of tered at each study visit (screen, baseline subcortical hyperintensities were localised and then weekly for 6 weeks). Data analyses in the frontal lobe area: 87% of lesions in The clinical outcome variables were re- the treatment-resistant group were in the Brain imaging procedures sponse (reduction in HRSD score of 50% frontal lobe, 79% of lesions in the non- All participants underwent brain magnetic or more) and remission (final HRSD score treatment-resistant group and 100% in resonance imaging using a 1.5 T (Signa, of 7 or less). We analysed the clinical data the control group, with no statistically General Electric, Milwaukee, Wisconsin, using the last observation carried forward significant difference between groups ww22 USA) whole-body imaging device. We ob- method. Group differences in demographic (( ¼0.825,0.825, PP¼0.22).0.22).

tained axial proton-density images, TT22-- and clinical variables involving continuous weighted images (echo time TE¼30/80 ms,30/80ms, data were computed using analysis of White-matter hyperintensities repetition time TR¼3000 ms, 25666192192 variance (ANOVA) (age) or unpaired tt-tests-tests and treatment outcome matrix; field of view 24 cm, flip angle 4588,, (HRSD scores, percentage change in HRSD After adjusting for age there was no statis- Nex value 0.5, 3 mm thick slices, no skip) scores).scores). tically significant relationship between the

181181

Downloaded from https://www.cambridge.org/core. 28 Sep 2021 at 17:13:13, subject to the Cambridge Core terms of use. IOSIFESCU ET AL

AUTHOR’ S P ROOF

Ta b l e 1 Clinical and demographic characteristics of the study participants total hyperintensity severity score and the clinical outcome measures of response to Major depressive disorderHealthy treatment and remission (Table 2). How- volunteervolunteer ever, after adjusting for age, subcortical Treatment- Non-treatment- group white-matter hyperintensities in the left resistant group resistant group ((nn¼35)35) hemisphere (and not in the right hemi- ((nn¼19) ((nn¼65) sphere) were correlated with lower rates of response (logistic regression coefficient Age, years: mean (s.d.)39.6 (7.6)40.7 (10.2) 39.2 (9.8)(9.8)39.2 2.31,2.31, ww22¼4.37,4.37, PP¼0.036, odds ratio 10.1, Gender, female: nn (%) 10 (53)24 (37)14 (40) 95% CI 1.16–87.71) and remission (logistic Cardiovascular risk score: mean (s.d.) 1.35 (1.22)(1.22)1.35 1.58 (1.21)(1.21)1.58 regression coefficient 2.69, ww22¼4.04,4.04, White-matter hyperintensity (WMH) score, (%)11 nn PP¼0.045, OR0.045,OR¼14.7, 95% CI 1.07–202.7) Total WMH after antidepressant treatment. When Total WMHWMHTotal ¼00 9 (47)9(47) 24 (37)(37)24 14 (40) treatment resistance status was used as a Total WMHWMHTotal ¼11 9 (47)9(47) 36 (55)(55)36 19 (54) stratification variable in the analysis, Total WMHWMHTotal ¼22 0 (0)0(0)4 (6)2(6) 2 (6) subcortical white-matter hyperintensities Total WMHWMHTotal ¼313 1 (5)(5) 1 (2)1(2) 0 (0)0(0) in the left hemisphere were significantly Any WMH (WMH551)1) 10 (53) 41 (63)(63)41 21 (60)(60)21 correlated with lower rates of treatment Severe WMH (WMH552)2) 1 (5)1(5) 5 (8)5(8)2(6) 2 (6) response (logistic regression coefficient 22 Periventricular WMH (PWMH) 772.46,2.46, ww ¼4.11,4.11, PP¼0.042, OR¼11.6,11.6, PWMHPWMH¼010 133 (68)(68) 36 (55)(55)36 20 (57)20(57) 95% CI 1.09–125.2), but the correlation PWMHPWMH¼11 6(32)6 (32) 27 27(42) (42)14 (40) between remission and subcortical hyper- intensities in the left hemisphere did not reach PWMHPWMH¼22 0 (0)0(0)2 (3)1(3) 1 (3) statistical significance (logistic regression PWMHPWMH¼33 0 (0)0(0) 0 (0)0(0) 0 (0)0(0) coefficientcoefficient 772.34,2.34, ww22¼3.19,3.19, PP¼0.074,0.074, Any PWHM (PWHM551) 6(32)6 (32) 29 29(45) (45)15 (43) OROR¼10.4, 95% CI 0.8–135.1). Of note, Severe PWMH (PWMH552)2) 0 (0)0(0)2 (3)1(3) 1 (3) there was no significant difference in the Deep (subcortical) WMH (DWMH) incidence of subcortical hyperintensities in DWMHDWMH¼00 11 (58)(58)11 46 (71)46(71) 22 (63)22(63) the left hemisphere between the treatment- DWMHDWMH¼11 7 (37)7(37) 16 (25)(25)16 12 (34)(34)12 resistant and the non-treatment-resistant DWMHDWMH¼22 0 (0)0(0)2 (3)1(3) 1 (3) depression groups (26% v.v. 23%;23%; PP440.05).0.05). DWMHDWMH¼313 1 (5)(5) 1 (2)1(2) 0 (0)0(0) After adjusting for age there was no statis- Any DWMH (DWMH551) 8 (42) 19 (29)(29)19 13 (37)(37)13 tically significant relationship between peri- Severe DWMH (DWMH552)2) 1(5)1 (5) 3(5) 1(3) ventricular white-matter hyperintensities and Left-sided DWMH 5 (26)5(26) 15 (23)(23)15 8(23) 8 (23) response to treatment or remission (Table 2). HRSD scores: mean (s.d.) Correlation between white-matter BaselineBaseline 21.6 (3.2)(3.2)21.6 20.2 (3.8)(3.8)20.2 hyperintensities and cardiovascular After treatment 18.4 (5.1)10.8 (7.3) risk factors Percentage change in score 0.14 (0.25) 0.48 (0.34)(0.34)0.48 The white-matter hyperintensity score Response, nn (%) 1(5)1 (5) 35 35(54) (54) was significantly correlated with the Remission,Remission, nn (%) 1 (5) 30 (46)30(46) cardiovascular risk score (PP¼0.037; Table HRSD,17-itemHRSD, 17-item Hamilton Rating Scale for Depression. 3). In a multiple ordinal logistic regression 1.Fazekas1. Fazekas score. analysis examining individual cardiovascular

Ta b l e 2 Multivariate logistic regression analyses for prediction of clinical response and remission by the severity of brain white-matter hyperintensities, adjusted for age

Lack of clinical response11 Lack of clinical remission22

Coefficient ww22 PP OR (95% CI)Coefficient ww22 PP OR (95% CI)

Total WMH 0.361.11 0.29 1.43 (0.74^2.76) 0.260.260.01 0.93 1.03 (0.55^1.91) Subcortical WMH 0.351.06 0.30 1.41 (0.73^2.71)0.13 0.140.140.71 1.13 (0.59^2.17) Left subcortical WMH 2.314.37 0.036* 10.07 (1.16^87.71)2.69 4.044.040.045* 14.71 (1.07^202.70) Right subcortical WMH 0.710.48 0.49 2.03 (0.27^15.21) 0.580.58 0.260.26 0.610.610.56 (0.06^5.20) Periventricular WMH 0.04 0.010.010.92 1.04 (0.46^2.37)0.03 0.00 0.990.991.00 (0.43^2.34)

OR, odds ratio; WMH, white-matter hyperintensity. **PP550.05. 1. Response defined as a reduction in the score on thethe17-item 17-item Hamilton Rating Scale for Depression of 50% or more. 2. Remission defined as final score of 7 or below on the Hamilton Rating Scale for Depression.

182182

Downloaded from https://www.cambridge.org/core. 28 Sep 2021 at 17:13:13, subject to the Cambridge Core terms of use. WHITE MATTER HYPERINTENSITIES AND TREATMENT OUTCOME IN DEPRESSION

AUTHOR’ S P ROOF 4.14)4.14) 6.05) 6.05) 75.17) 75.17) 30.93) 59.58)59.58) 30.93) 429.25) 292.95)292.95) 429.25) 11 2.23 (1.21^ 0.91 (0.01^ 0.42 (0.03^ 21.92 (1.12^ 24.61 (2.07^ OR (95% CI) OR (95% CI) PP .6 54 04^.0(0.49^ 0.16 5.40 (0.49^5.40 0.96 0.91 (0.01^ 0.52 0.42 (0.03^ .7 33 03^.9(0.37^ 0.27 3.39 (0.37^3.39 0.041* 21.92 (1.12^ 0.011* 2.23 (1.21^ 22 Severe WMH ww 1.17 0.27 4.14 0.041* 0.02 0.96 .3 002002 24.61 (2.07^ 6.43 0.012*0.012* 6.53 0.011* 0.40 0.52 Logistic regression 0.92 0.92 0.02 0.86 0.86 0.40 1.69 0.16 1.901.90 1.22 1.17 3.09 4.14 0.80 3.20 6.43 0.80 6.53 Coef. Coef. 77 77 P 0.59 0.89 0.29 1.69 0.017* 1.22 Z ZP 0.14 0.89 1.71 3.09 0.0880.088 1.06 0.29 2.39 0.017* 2.28 3.20 0.023*0.023* 0.140.14 0.14 1.22 .4.42.28 1.441.44 0.53 0.64 0.300.30 0.530.53 0.59 1.22 2.39 0.53 1.71 0.64 1.06 Coef.Coef. Ordinal logistic regression Logistic regression P 0.008* Periventricular WMH Severe WMH F FP 7.41 0.008* ANOVA Ordinal logistic regression d.f.d.f. 1,79 1,79 7.41 P 0.31 0.078 0.94 0.88 0.48 0.14 0.88 Z ZP 1.76 0.078 .6.60.94 0.060.06 0.28 0.770.77 1.01 0.31 77 0.15 0.140.14 0.14 0.15 0.28 0.63 0.04 0.64 0.45 0.63 1.76 0.04 0.64 1.01 0.45 0.48 0.700.70 Coef.Coef. 77 Ordinal logistic regression ANOVA P Subcortical WMH Periventricular WMH F FP 4.06 0.047*0.047* ANOVA Ordinal logistic regression d.f.d.f. 1,79 1,79 4.06 P 0.053 0.016* 0.99 0.001* Z ZP 1.94 0.053 1.65 0.0980.098 0.62 0.530.53 3.20 0.001* 2.40 0.016* 0.01 0.99 0.970.97 1.94 0.03 0.990.99 1.65 1.56 0.57 0.03 0.01 0.57 0.62 1.56 2.40 1.01 1.0 1 3.20 Coef. Coef. Ordinal logistic regression ANOVA Total WMH Subcortical WMH P F FP 4.50 0.037*0.037* ANOVA ANOVA Ordinal logistic regression d.f.d.f. 1,79 1,79 4.50 45,45, 55 2.2. 55 orltosbtenwiemte yeitniisadcrivsua risk and factors cardiovascular between hyperintensities white-matter Correlations orltosbtenwiemte yeitniisadcrivsua risk and factors cardiovascular between hyperintensities white-matter Correlations 55)55) 55 0.05.0.05. 55 Hypercholesterolaemia Diabetes mellitus Smoker Family history of disease cardiovascular Hypertension Total cardiovascular risk Age/gender (male female Ta b l e 3 Risk factor analysis ofANOVA, variance; coef., WMH, coefficient; hyperintensity. white-matter PP ** 1. F azeka s score Hypercholesterolaemia Diabetes mellitus Smoker Family history of disease cardiovascular Hypertension Total cardiovascular risk Age/gender (male female ANOVA, analysis of variance; coef., coefficient; WMH, white-matter hyperintensity. 1. F azeka s score Risk factor Total WMH Ta b l e 3

183183

Downloaded from https://www.cambridge.org/core. 28 Sep 2021 at 17:13:13, subject to the Cambridge Core terms of use. IOSIFESCU ET AL

AUTHOR’ S P ROOF

risk factors as predictors, greater total brain of Greenwald et aletal (1998), in which left hyperintensities of specific white-matter white-matter hyperintensity score was frontal subcortical white-matter lesions sig- tracts involved in mood regulation, may associated with age (PP¼0.016) and with nificantly correlated with the diagnosis of explain our observed result of a selective hypertension (PP¼0.001). Two other major depression. Also, previous reports impact of left-hemisphere subcortical cardiovascular risk factors did not reach in post-stroke depression have indicated hyperintensitieshyperintensities on treatment outcome. statistical significance in relation to total left-sided frontal lobe lesions as predictors Other predictorsof treatment outcome, hyperintensity score: cholesterol level of severity of depression at 3 months and which may mediate the relationship ((PP¼0.053) and family history of cardiovas- 6 months after the stroke (Robinson etet between white-matter hyperintensity and cular disease (PP¼0.098). The presence of alal, 1985); left frontal localisation of stroke treatment response, might have been missed diabetes and smoking status were not corre- was not predictive of depression at long- in this study owing to the relatively small lated with the total hyperintensity score. term follow-up (Shimoda & Robinson, sample size. The presence of severe hyperintensities also 1999). However, this pattern would be con- There are several limitations to our correlated with age (PP¼0.012) and with sistent with the results we are reporting study. First, we used a whole-brain rating hypertension (PP¼0.041), as well as with here, as only short-term, acute-phase treat- scale (Fazekas et aletal, 1987) to assess the the total cardiovascular risk score ment outcomes were measured in our severity of brain white-matter hyper- ((PP¼0.011).0.011). study.study. intensities. Although most studies on this The total cardiovascular risk score was A second important finding in our topic reported using modified versions of correlated with the severity of periventricu- study was that brain white-matter hyper- the Fazekas scale, this method does not lar hyperintensities (PP¼0.008) and of sub- intensities were correlated with cardio- allow for a detailed morphological and cortical hyperintensities (PP¼0.047). In a vascular risk factors in non-elderly people volumetric analysis of brain white-matter multiple logistic regression, the severity of with major depressive disorder. It is there- hyperintensity (Taylor et aletal, 2003). We periventricular hyperintensity correlated fore possible that a large proportion of did not measure hyperintensity localisa- with age ((withage PP¼0.023) and hypercholestero- brain hyperintensities in this younger popu- tions such as , which have laemia (laemia(PP¼0.017), whereas hypertension lation might be vascular in origin. Our been previously described as associated did not reach statistical significance result is consistent with findings in non- with treatment outcome in major depres- ((PP¼0.088). Hypertension also did not reach psychiatric populations, where brain sive disorder (Simpson et aletal, 1998). Second, statistical significance in relation to sub- white-matter hyperintensities have been we measured only total cholesterol and not cortical hypertensities (PP¼0.078). Other associated with age, cerebrovascular dis- fractions of cholesterol, therefore our re- cardiovascular risk factors did not correlate ease (Awad et aletal, 1986; Fazekas et aletal,, sults may not reflect the full impact of this with the severity of periventricular or sub- 1993) and cardiovascular risk factors such cardiovascular risk factor. Third, we have cortical white-matter hyperintensities. as smoking, arterial hypertension and in- a potential sampling bias, as we enrolled creased serum cholesterol (Breteler et aletal,, participants from two antidepressant trials DISCUSSION 1994; Liao et aletal, 1997; Schmidt et aletal,, with specific inclusion and exclusion 1997). In neuropathological analyses of criteria and with different treatments This is to our knowledge the first study brains from people with major depressive (fluoxetine and nortriptyline); as a result, investigating the impact of brain white- disorder, subcortical white-matter hyperin- this sample may not directly reflect the matter hyperintensities on antidepressant tensities were all ischaemic (Thomas et aletal,, typical out-patient population. treatment outcome in a non-geriatric study 2002), whereas periventricular lesions had sample. There are several important find- multiple causes (Thomas et aletal, 2003).,2003). ACKNOWLEDGEMENTS ings in this study. First, although we found One possible interpretation of our re- no correlation between global white-matter sults is that cardiovascular risk factors cor- The study was supported by a National Alliance for hyperintensity measures and clinical out- relate with a higher severity of subcortical Research on Schizophrenia and Depression come, subcortical hyperintensities in the white-matter lesions, which in turn corre- (NARSAD) Young Investigator Award (D.V.I.), by a left hemisphere were associated with poor lates with poor treatment outcome in Kaplen Fellowship Award for Depression from Harvard Medical School (D.V.I.) and by National outcome of antidepressant treatment. Pre- depression. This interpretation would be Institute of Mental Health grants R01-MH48483 vious studies in geriatric populations have consistent with the ‘vascular depression’ (M.F.)and(M.F.) and MH58681 (P.F.R.). described an association between sub- model. However, the vascular depression cortical and basal ganglia white-matter hypothesis would call for a higher preva- REFERENCES hyperintensity (and not periventricular lence of hyperintensities in patients with de- hyperintensity) and poor treatment out- pression compared with normal controls, Awad, I. A., Johnson, P.C., Spetzler,R.Spetzler, R. F., et aletal (19 8 6) come in major depressive disorder (Hickie which was not found in our study. Also, Incidental subcortical lesions identified on magnetic et aletal, 1997; O’Brien et aletal, 1998; Simpson the vascular depression model would sug- resonance imaging in the elderly.II: postmortem pathological correlations. Stroke,, 1717,1090^1097. et aletal, 1998; Steffens et aletal, 2002). However, gest a relationship between global measures to our knowledge there is no other of brain white-matter hyperintensities and Breteler, M. M., van Swieten, J. C., Bots, M. L., et aletal report indicating a differential impact of treatment outcome, whereas in our sample (19 94) Cerebral white matter lesions, vascular risk factors, and cognitive function in a population-based left-hemisphere subcortical white-matter only subcortical hyperintensities in the left study: the Rotterdam Study. NeurologyNeurology,, 4444,,1246^1252. 1246^1252. hyperintensity on treatment outcome in hemisphere (and no other hyperintensities) major depressive disorder. Further studies were associated with poor response to Coffey, C. E.,Wilkinson,W. E.,Weiner, R. D., et al (19 93) Quantitative cerebral anatomy in depression. A will be needed to validate this association. antidepressant treatment. Other factors, controlled magnetic resonance imaging study. Archives ofofArchives Our finding is consistent with the report such as interruption by white-matter General Psychiatry,, 50,,7^16. 7^16.

184184

Downloaded from https://www.cambridge.org/core. 28 Sep 2021 at 17:13:13, subject to the Cambridge Core terms of use. WHITE MATTER HYPERINTENSITIES AND TREATMENT OUTCOME IN DEPRESSION

AUTHOR’ S P ROOF

de Groot, J. C., de Leeuw, F. E., Oudkerk, M., et aletal (2000) Cerebral white matter lesions and depressive symptoms in elderly adults. Archives of General Psychiatry,, CLINICAL IMPLICATIONS 57,1071^1076. && Expert Panel on Detection, Evaluation and Subcortical white-matter hyperintensities in the left hemisphere were associated Treatment of High Blood Cholesterol in Adults with poor response to antidepressant treatmentin major depressive disorder,but we (2001)(2001) Executive summary of theThird Report of the National Cholesterol Education Program (NCEP) found no correlation between global white-matter hyperintensity measures and Expert Panel on Detection, Evaluation and Treatment of clinical outcome. High Blood Cholesterol in adults (Adult Treatment Panel III).III).Panel JAMAJAMA,, 285285, 2486^2497. && In younger people with major depressive disorder, the presence of such Fava, M., Alpert, J., Nierenberg, A. A., et aletal hyperintensities was associated with cardiovascular risk factors (hypertension and (2002)(2002) Double-blind study of high-dose fluoxetine age) and with the total cardiovascular risk score. versus lithium or desipramine augmentation of fluoxetine in partial responders and nonresponders to && fluoxetine. Journal of Clinical Psychopharmacology,, 2222,, We found no difference in the prevalence or distribution of white-matter 379^387. hyperintensities between healthy volunteers and two groups of younger people with

Fazekas, F., Chawluk, J. B., Alavi, A., et aletal (19 87) MR depression (treatment-resistant and non-treatment-resistant). signal abnormalities at 1.5 T in Alzheimer’s dementia and normal aging. American Journal of Roentgenology,, 149,, LIMITATIONS 351^356.351^356. && Fazekas, F., Kleinert, R.,Offenbacher,H., et al (19 93) We cannot exclude significant effects of white-matter hyperintensities in specific Pathologic correlates of incidental MRI white matter brain regions not examined here. signal hyperintensities. Neurology,, 43,1683^1689. && Folstein, M. R., Folstein, S. E. & McHugh, P.R.(19 (1975) 75) We cannot assess the role of fractions of cholesterol (important cardiovascular ‘Mini-Mental State’: a practical method of grading the risk factors), as only total cholesterol was measured. cognitive state of patients for the clinician. Journal ofofJournal Psychiatric Research,, 1212,189^198.,189^198. && We enrolled participants from two separate antidepressant trials, using different Greenwald, B. S., Kramer-Ginsberg, E., Krishnan, treatments (fluoxetine and nortriptyline). K. R., et aletal (19 9 8)8)(19 Neuroanatomic localization of magnetic resonance imaging signal hyperintensities in geriatric depression. Stroke,, 29, 613^617.,613^617.

Hamilton, M.(1967) Development of a rating scale for primary depressive illness. British Journal of Social and Clinical Psychology,, 66, 278^296.,278^296. DAN V.IOSIFESCU,MD,Depression Clinical and Research Program, Massachusetts General Hospital, PERRYF. Hickie, I., Scott, E.,Wilhelm, K., et aletal (19 9 7) RENSHAW,RENSHAW,MD,PhD,IN MD, PhD, INKYOON KYOON LYOO,LYOO,MD,PhD,Brain MD, PhD,Brain Imaging Center,Center,McLean McLean Hospital, Harvard Medical Subcortical hyperintensities on magnetic resonance School, Boston, Massachusetts,USA; HO KYU LEE, MD,PhD,Department of Radiology,Sung Kyun Kwan imaging in patients with severe depression ^ a University,Seoul,Korea; ROY H. PERLIS, MD,GEORGE I. PAPAKOSTAS, MD, ANDREW A. NIERENBERG, MD, longitudinal evaluation. Biological Psychiatry,, 42,, MAURIZIO FAVA, MD,Depression Clinical and Research Program, Massachusetts General Hospital, Harvard 367^374. Medical School, Boston, Massachusetts,USA Krishnan, K. R., Hays, J. C. & Blazer, D. G.(19 (1997) 9 7) MRI-defined vascular depression. American Journal of Correspondence: Dr DanV.Iosifescu,Massachusetts General Hospital, 50 Staniford Street, suite 401, Psychiatry,, 154, 497^501. Boston, Massachusetts 02114,USA.Tel: +1 617 724 7741; fax: +1 617 724 3028; e-mail: @@ Lenze, E., Cross, D., McKeel, D., et aletal (19(1999) 9 9) White diosifescu partners.org matter hyperintensities and gray matter lesions in physically healthy depressed subjects. American Journal (First received 29 September 2004, final revision 14 February 2005, accepted 18 February 2005) of Psychiatry,, 156156,1602^1607.

Liao, D., Cooper, L., Cai, J., et aletal (19 9 7) The Schmidt, R., Fazekas, F., Hayn, M., et aletal (19 9 7) RiskRisk Taylor,W.Taylor,W.D., D., MacFall, J. R., Steffens, D. C., et aletal prevalence and severity of white matter lesions, factors for microangiopathy-related cerebral damage in (2003)(2003) Localization of age-associated white matter their relationship with age, ethnicity, gender and the Austrian stroke prevention study. Journal of the hyperintensities in late-life depression. Progress in cardiovascular disease risk factors: the ARIC Study. Neurological Sciences,, 152152,15^21. Neuropsychopharmacological and Biological Psychiatry,, 2727,, Neuroepidemiology,, 1616,149^162.,149^162. 539^544.539^544. Shimoda, K. & Robinson, R. G.(19 (1999) 9 9) TheThe Lyoo, I. K., Lee, H. K., Jung, J. H., et aletal (2002)(2002) WhiteWhite relationship between poststroke depression and lesion Thomas, A. J., O’Brien, J. T., Davis, S., et aletal (2002)(2002) matter hyperintensities on magnetic resonance imaging location in long-term follow-up. Biological Psychiatry,, 4545,, Ischemic basis for deep white matter hyperintensities in of the brain in children with psychiatric disorders. 187^192.187^192. major depression: a neuropathological study. Archives ofofArchives Comprehensive Psychiatry,, 4343, 361^368.,361^368. General Psychiatry,, 5959, 785^792. Simpson, S., Baldwin, R. C., Jackson, A., et aletal (19 9 8) Nierenberg, A. A., Papakostas, G. I., Petersen, T., Is subcortical disease associated with a poor response to Thomas, A. J., O’Brien, J.T.,J. T., Barber, R., et al (2003)(2003) et aletal (2003)(2003) Nortriptyline for treatment-resistant antidepressants? Neurological, neuropsychological and A neuropathological study of periventricular white depression. Journal of Clinical Psychiatry,, 6464, 35^39.,35^39. neuroradiological findings in late-life depression. matter hyperintensities in major depression. Journal of Psychological Medicine,, 28,,1015^1026. 1015^1026. Affective Disorders,, 76, 49^54.,49^54. O’Brien, J., Ames, D., Chiu, E., et aletal (19 9 8)8)(19 Severe deep white matter lesions and outcome in elderly Spitzer, R. L.,Williams, J. B.W., Gibbon, M., et aletal Yanai, I., Fujikawa, T., Horiguchi, J., et aletal (19 9 8)8)(19 The patients with major depressive disorder: follow up study. (19 8 9) Structured Clinical Interview for DSM ^ III ^ R Patient 3-year course and outcome of patients with major BMJBMJ,, 317317,,982^984. 982^984. edition (SCID ^ P). New York: Biometrics Research depression and silent cerebral infarction. Journal ofofJournal Department, New York State Psychiatric Institute. Affective Disorders,, 47,25^30. Robinson, R. G., Starr, L. B., Lipsey, J. R., et aletal (19 8 5) A two-year longitudinal study of poststroke mood Steffens, D. C., Krishnan, K. R., Crump, C., et aletal Wilson, P.W., D’Agostino, R. B., Levy, D., et aletal disorders. In-hospital prognostic factors associated with (2002)(2002) Cerebrovascular disease and evolution of (19 9 8) Prediction of coronary heart disease six-month outcome. Journal of Nervous and Mental depressive symptoms in the cardiovascular health study. using risk factor categories. CirculationCirculation,, 9797,, Disease,, 173, 221^226.,221^226. Stroke,, 33,1636^1644. 1837^1847.1837^18 47.

185185

Downloaded from https://www.cambridge.org/core. 28 Sep 2021 at 17:13:13, subject to the Cambridge Core terms of use.