June 02, 2021
Whole-body imaging that is performed routinely in the treatment of patients with multiple myeloma could offer the added benefit of flagging those with possible depression, "without extra dedicated brain images or increased radiation exposure or cost," according to researchers at the University of Arkansas, in Little Rock, Arkansas.
The team analyzed a series of fludeoxyglucose F-18 positron-emission tomography–CT (FDG PET/CT) scans taken for staging, to assess treatment response, and for surveillance for patients with multiple myeloma. They concluded that the images could also pick up brain metabolic changes suggestive of major depression.
"This would not replace the need for a psychological evaluation," they emphasize. "However, this may alert the clinician to metabolic changes in the brain that have been associated with depression," reported Xiaofei Wang, MD, PhD, and colleagues from the Department of Radiology at the University of Arkansas for Medical Sciences.
"The patient may have sub-clinical depression or may have been embarrassed to express their symptoms to their oncologist. This would serve as a supplemental screening exam to the more traditional questionnaires," they add.
"If you could pick up an incidental finding [of depression] on a scan that was already happening, that would be useful, that would be unique ? but I don't think their study shows that," she told Medscape Medical News.
The study was published online on May 6 in PLOS ONE.
The study used brain images extracted from whole-body FDG PET/CT scans of 134 patients with multiple myeloma; 38 of the patients had been diagnosed with major depression, and the remaining 96 had not. Statistic parameter mapping showed that there was significantly lower global FDG activity in the brains of patients with depression than in those of patients without depression (P < .001), indicating "less glutamate neurotransmission through glucose oxidation process, which links to major depression," the authors reported.
In addition to the global brain changes, the analysis also identified clusters of metabolic differences between the depressed and nondepressed patients. Specifically, it showed five significant hypometabolic clusters and three significant hypermetabolic clusters in the group with depression. "Interestingly, hyper-metabolic clusters were anatomically localized in limbic system bilaterally and right brainstem, whereas hypo-metabolic clusters predominantly involved bilateral frontal, parietal, and right temporal lobes," the researchers note.
They then applied machine learning to this pattern of clusters to create two predictive models for major depression screening. The models were found to have 85.1% accuracy, 79% sensitivity, and 88% specificity for detecting depression.
"With the rapid advancement of AI [artificial intelligence], there is potential for automated screening of patients during the staging PET/CT for major depression," they suggest. "This may increase awareness and identification of patients needing additional support for depression."
The researchers note that a limitation of the study is that "potential confounding variables, such as current treatment regimens, coexisting anxiety disorders or other comorbidities, and depressive symptoms at the time of the scan were not taken into account and represent possible biases."
Li, who developed and leads the PMCC's Distress Assessment and Response Tool program, said her main concern with the study, aside from the small sample size, is the how the diagnosis of depression was established.
"The diagnosis of depression in cancer is actually very challenging ? it's really hard to distinguish the symptoms of depression from the symptoms of cancer, and I have no idea from the paper what the diagnosis of depression was based on," she said. "They're saying they can incidentally pick up ? maybe depression ? but because we know nothing about the state of those patients, we don't know if they're actually picking up depression or something else ? perhaps a predisposition to depression," she explained.
In closing, Li, who also conducts research with machine learning and blood biomarker development for depression for patients with cancer, said, "The most I could say is this is a proof of principle ? that you might be able to use these machine learning algorithms to differentiate something (although I'm not sure what they're differentiating), but to say it's a screening tool for depression in cancer is so overreaching I can't even describe how offensive that is."
The study received no specific funding. The authors and Li have disclosed no relevant financial relationships.
Health Solutions From Our Sponsors