White matter hyperintensity detected by magnetic resonance imaging and lithium response in bipolar disorder: A preliminary observation


Correspondence address: TadafumiKatoMD Department of Neuropsychiatry, Faculty of Medicine, University of Tokyo, Hongo 7-3-1, Bunkyo, Tokyo, 113-8655, Japan. Email: tadafumi-tky@umin.ac.jp


The possible relationship between deep white matter hyperintensity (WMHI) lesions detected by magnetic resonance imaging and response to lithium was examined in 16 patients with bipolar disorder who had been under maintenance treatment with lithium for more than 1 year. Bipolar patients who had higher scores of WMHI responded significantly better to lithium (r = 0.57, P < 0.05) than did those who had lower scores. This preliminary result suggests that the presence of WMHI may be associated with a better response to lithium.


Bipolar disorder is a socially disabling mental disorder that affects approximately 1% of the population. Although there are now several available mood stabilizers, 1 lithium is still the drug of first choice for maintenance treatment for bipolar disorder. 2 However, lithium is not effective for all bipolar patients. Lithium can completely prevent recurrence in only 50% of patients with bipolar disorder, while other patients are completely resistant to lithium treatment or show only a partial response. 3 Although it takes many months, or even several years, to determine whether a patient is a responder or not, there are as yet no established laboratory tests or biological markers that can predict lithium response. 3

It has been reported that patients with bipolar disorder more frequently have subcortical hyperintensity (SCH) lesions in the brain detected by T2-weighted magnetic resonance imaging (MRI) than do age-matched healthy volunteers. 4,5 Subcortical hyperintensity in patients with bipolar disorder is not related to cardiac risk factors 6,7 but to positive family history, which suggests it may not be attributable to merely ischemic lesions. 8 Although its histopathological basis has never been reported, our recent study using phosphorus-31 magnetic resonance spectroscopy suggested that SCH in subjects with bipolar disorder might have a different biochemical basis to that in healthy aged subjects. 9 While SCH is related to poor treatment response in major depression, 10 the clinical significance of SCH in bipolar disorder is not yet well known. 11 To the best of our knowledge, there is no controlled study in which the efficacy of lithium was compared between bipolar patients with and without organic factors. 11

Subcortical hyperintensity includes periventricular hyperintensity (PVH), subcortical gray matter hyperintensity (SGH) and deep white matter hyperintensity (WMHI). 4 The pathogenetic role of each of the SCH in bipolar disorder has not as yet been clarified.

The goal of this preliminary study was to examine whether SCH is associated with response to maintenance treatment with lithium.


The subjects comprised 16 patients with bipolar disorder (12 females and four males) who had been under maintenance treatment with lithium at the Shiga University of Medical Science Hospital and Clinic for more than 1 year. Their serum lithium concentrations were maintained at 0.3–1.0 mmol/L. Their ages were 47.1 ± 15.1 (mean ± SD) years and their DSM-IV diagnoses were bipolar I disorder (n = 10) or bipolar II disorder (n = 6).


Assessment of response to maintenance lithium treatment

During hospitalization the patients had been evaluated using either the Hamilton Rating Scale for Depression or the Petterson Rating Scale for Mania. Although such instruments had not been used in the out-patient clinic, the mood state and compliance of the patients had been clinically evaluated by senior psychiatrists every 2 weeks. Response to maintenance lithium treatment was assessed by reviewing these chart records at the hospital and out-patient clinic. The patients were classified into three groups with regard to lithium response. Complete non-responders (Class 0, n = 7) continuously relapsed after initiation of lithium treatment despite their good compliance. Partial responders (Class 1, n = 6) had relapses requiring additional treatment while their compliance was good; however, the severity or frequency of episodes was decreased compared to before lithium treatment. Complete responders (Class 2, n = 3) had no relapses that required additional treatment while their compliance was good. These clinical evaluations were performed blindly to MRI data. The observation period was 4.3 ± 2.5 years. There was no significant difference of observation period among these three groups (3.9 ± 2.5, 5.6 ± 2.2 and 2.4 ± 0.8 years, respectively) (F = 2.2, P = 0.14 using one-way analysis of variance).

Magnetic resonance imaging

Magnetic resonance images were obtained using a SIGNA 1.5 Tesla MR system (GE Medical Systems, Milwaukee, WI, USA). T2-weighted and proton density-weighted double-spin echo axial images were taken with a field of view (FOV) of 24 cm, a slice thickness of 5 mm with an interscan gap of 2.5 mm, a 256 × 256 matrix, a repetition time (TR) of 2000 msec, echo times (TE) of 80 msec and 20 msec, respectively, and a number of excitations (NEX) of 2. T1-weighted axial images were also obtained with a TR of 500 msec and a TE of 20 msec.

These MRI films were first qualitatively inspected by a neuroradiologist, then quantitatively scored by two neuropsychiatrists. White matter hyperintensity, PVH, and SGH were scored by T2-weighted magnetic resonance images and proton density-weighted images using the four-point scale by Coffey et al.12 White matter hyperintensity was rated as follows: 0 = absent; 1 = punctate foci; 2 = beginning confluence of foci; and 3 = large confluent areas. Periventricular hyperintensity was graded as 0 = absent, 1 = caps or pencil-thin lining, 2 = smooth halo, and 3 = irregular PVH extending into the deep white matter. Subcortical gray matter hyperintensity was graded as 0 = absent, 1 = punctate, 2 = multipunctate, 3 = diffuse. Predefined visual standards were prepared for comparison with the MRI films in these patients.


Bipolar patients who had higher scores of WMHI responded to lithium significantly better (r = 0.57, P < 0.05 using Spearman’s rank-order correlation) than did those with lower scores ( Fig. 1). However, there was no significant relationship between lithium response and PVH (r = 0.29, P = 0.27) or SGH (r = 0.12, P = 0.65).

Figure 1.

Relationship between white matter hyperintensity scores by T2-weighted magnetic resonance imaging and lithium response in patients with bipolar disorder. A statistically significant correlation was found (r = 0.57, P < 0.05 using Spearman’s rank-order correlation). WMHI, white matter hyperintensity.

Patients presenting as WMHI positive (WMHI score > 0) tended to have better scores of lithium response (n = 8, P = 0.08 by Mann–Whitney U-test) than did those presenting as WMHI negative (n = 8). The former group was significantly older and tended to have a higher age at onset compared with the latter group ( Table 1).

Table 1.  Characteristics of subjects with bipolar disorders
 Deep white matter
  1. *P < 0.05; +P < 0.10 using Mann–Whitney U-test.

  2. NS, non-significant; BPI, bipolar I disorder; BPII, bipolar II disorder.

Sex (female/male)5/37/1NS
Subtype (BPI/BPII)6/24/4NS
Family history3/55/3NS
Age (mean + SD)38.5 + 15.155.8 + 9.7*
Age at onset33.8 + 15.947.0 + 11.0+
(mean + SD)
Duration of illness5.5 + 6.57.0 + 4.1NS
(mean + SD)


These preliminary results suggest that the existence of WMHI, but not of PVH and SGH, may be associated with a better response to maintenance lithium treatment. Although the present finding may seem contradictory to previous studies that indicate an association between the presence of WMHI and poor treatment response in patients with major depression, 10 the present finding may not be so surprising considering that both WMHI 8 and favorable lithium response were associated with positive family history in bipolar disorder. 3 The present finding is compatible with previous studies suggesting that lithium was also effective for treating patients with secondary mania due to organic brain syndrome, 13–16 although such patients may be more susceptible to side effects. 13

It should be noted that the effects of the confounding factors, that is, family history, diagnostic subtype, age, age at onset, clinical symptoms, episode frequency, and duration of illness, could not be ruled out in this study. A further study controlling these confounding variables in a larger patient population is needed to confirm this preliminary finding. A cohort study examining the effects of the presence of WMHI on long-term outcome of maintenance treatment with lithium would be the best way to prove a relationship between WMHI and lithium response.

In this preliminary study, the location of WMHI lesions was not considered, except for the finding that PVH did not correlate with lithium response. Dupont et al. suggested that WMHI in the frontal region rather than in the posterior region may be correlated with bipolar disorder. 8 This needs further clarification.

In spite of these limitations, this is the first study, to the best of our knowledge, to suggest a possible relationship between WMHI and favorable lithium response in bipolar disorder. If this finding is confirmed, it may be useful in predicting treatment response before long-term maintenance treatment in this disorder.