General:
Figure 1 illustrates the distribution of benign intracranial lesions and malignant brain tumors. Benign lesions included meningiomas, low-grade gliomas, pituitary tumors and cavernous hemangiomas. Malignant brain tumors only included glioblastomas and brain metastases.
Three lesions were not able to be defined histologically or were identified as inflammatory processes. These patients were excluded from statistical analysis.
We recruited 61 patients for this study, their age ranged from 22 and 78 (57.2±13.81) years. 57% were female, 43% were male. 3 Patients were excluded after pathological diagnosis failed to show a tumor.
Only 32 patient of the 61 patients initially included (52%) completed the study. Not all of those 32 patients completed all questionnaires at each interview. The 29 patients that discontinued the study did so for the following reasons: impairment (10/29), missed follow-up (16/29), secondary exclusion if histological workup did not diagnose a tumor (3/29).
Only 50% of the patients with GBM and 40% of patients diagnosed with brain metastasis completed all 6 interviews of the study. Whereas 69% of the patients with benign lesions were able to complete all questionnaires until final follow-up. For further analysis we than stratified our patient cohort into two groups and combined patients with GBM and brain metastasis into one group of patients diagnosed with a malignant brain tumor compared to patients diagnosed with benign brain tumors.
Only one patient showed a positive screening test for social isolation on Lubben social network scale.
No correlation could be identified between sex and depression, intelligence scores and depression, as well as elevated risk for social isolation and depression.
There was a significant positive correlation between scores on CES-D and physical disability on MRS (preoperative correlation: 0.331, p = 0.015; follow-up correlation: 0.483, p = 0.013).
Established depression scores:
Prior to surgery and therefore before a diagnosis was made and communicated to the patient, scores on depression scales were similar for patients with malignant and benign lesions. CES-D, Beck depression inventory and PHQ-9 showed positive correlation before surgery (CES-D and PHQ-9: 0.762 (p < 0.001); CES-D and BDI-II: 0.774 (p < 0.001); PHQ-9 and BDI-II: 0.758 (p < 0.001)).
Prior to surgery the mean CES-D score in patients with benign tumors was 12.9 (±10.5) points with 11/32 (34%) patients having a score of 16 points or more suggesting clinically relevant depression. The mean score in patients with malignant brain tumors was 12.3 (±8.2) with the score adding up to 16 or more points in 9/25 patients (36%), respectively.
Figure 2 shows that CES-D scores increased significantly for patients with malignant brain tumors (p=0.029) and decreased significantly for patients with benign brain tumors (p=0.049) over time. Further, it could be shown that CES-D scores deviated significantly (p<0.001) over time in direct comparison between patients with benign and malignant brain tumors.
Similar results were seen for BDI tests, as presented in Figure 3. The interaction between the patient group with benign tumors and the patient group with malignant tumors showed deviation with p < 0.0001.
When comparing patients with benign tumors to patients with GBM (Metastases excluded), no significant trend towards increase or decrease in the depression scales could be shown for GBM alone. However, 21 days after surgery, 78.5% of patients with GBM showed pathological results on CES-D compared to 22% of patients with benign brain lesions, whilst 75% of patients with GBM showed pathological results on CES-D at follow-up, compared to 18% of patients with benign brain lesions (p = 0.002 and p = 0.007, respectively). (Table 1)
Table 1: Number of patients with CES-D scores of 16 points or more
|
Time of testing
|
Benign (%)
|
GBM (%)
|
Test
|
p-value
|
Pre-OP
|
11 (34%**)
|
5 (31%**)
|
Chi2
|
0.829
|
7 days post-OP
|
5 (22%**)
|
5 (56%**)
|
Fisher
|
0.096
|
21 days post-OP
|
5 (22%**)
|
7 (88%**)
|
Fisher
|
0.002*
|
35 days post-OP
|
5 (25%**)
|
4 (67%**)
|
Fisher
|
0.138
|
Follow-up
|
4 (18%)
|
6 (75%)
|
Fisher
|
0.007*
|
Table 1: Values with p < 0.05 marked with *; ** = rounded to full percent. Post-OP = Post-Operation
Study-specific questionnaire (SSQ):
Using the SSQ, there was a non-significant differential trend of patients diagnosed with GBM for the following items: Patients diagnosed with GBM tended towards a higher level of sadness, were more afraid to lose control, had less plans for the future, felt more powerless, had more changes in their social life, and described themselves as less emotionally stable when compared to patients with benign lesions or brain metastases.
None of the changes, however, were significant.
However, the results showed tendencies towards differing results in patients with metastases and patients with GBM (as shown exemplarily in Figure 4).
To validate, whether certain questions on the SSQ were suitable for detecting depression, we focused on correlation coefficients between the single items of the SSQ (Supplemental Material 1) and CES-D scoring results. We found correlations for the following questions on the questionnaire: 1, 2, 4, 7, 9, 11, 13, 14, 17, and 19 (p < 0.05).
As an example, patients who strongly agreed with the statement: “I feel helpless”, were more likely to score higher points in CES-D (correlation coefficient 0.477; p = 0.0003). Patients who gave an answer closer to “yes” when asked, if they could cope with their diagnosis were much less likely to score high on CES-D (correlation coefficient -0,581; p < 0.0001).
In some cases, correlations and weak correlations could be identified between CES-D and the study specific questionnaire (SSQ).
A representative trend within the SSQ is illustrated in Figure 4.
In addition to correlation with CES-D and BDI-II, the study-specific questionnaire was also tested for internal consistency using the Cronbach´s alpha. Again, questions 1, 2, 4, 7, 9, 10, 11, 13, 14, 17 and 19 were included in the analysis. Questions 2, 4 and 13 matched least with the other questions (Table 2).
Table 2: Cronbach’s alpha standardized variables tested as internal consistency
|
|
SSQ #1
|
SSQ #2
|
SSQ #4
|
SSQ #7
|
SSQ #9
|
SSQ #10
|
SSQ #11
|
Cronbach’s alpha
|
0.27
|
0.04
|
-0.05
|
0.14
|
0.28
|
0.28
|
0.18
|
|
SSQ #13
|
SSQ #14
|
SSQ #17
|
SSQ #19
|
Cronbach’s alpha
|
0.09
|
0.3
|
0.16
|
0.23
|
Number needed to screen
Exemplarily, assuming a sensitivity of 85% for depression screening using CES-D [11] and a prevalence of 90% for patients with malignant brain tumors [2], a number needed to screen for detection of depression in patients with GBM of 1.59 with a false positive screening rate of 2.8%. was calculated (Table 3).
Table 3: Cross table based on 1000 Patients with a prevalence of 90% for depression
|
|
Detected by Questionnaires
|
|
|
Detected by family physician
|
|
with
depression
|
without depression
|
|
|
with
depression
|
without depression
|
Positive testing
|
765
|
28
|
793
|
Positive testing
|
135
|
0
|
Negative testing
|
135
|
72
|
207
|
Negative testing
|
765
|
100
|
|
900
|
100
|
1000
|
|
900
|
100
|
Table 3: This table is relevant for those patients with worrisome screening results that are not identified by their family physician. This applies for 630 of 1000 patients, concluding in a number needed to screen of 1.59 (reciprocal of 630/1000)