This study indicates that some schizophrenia patients are undergoing a process of progressive brain change, defined as a decrement in brain tissue volume that is occurring at a more rapid rate in patients than in control subjects. Thus, schizophrenia has a neuro-progressive component, defined as tissue volume decrease occurring after onset. The changes include both GM and WM and manifest their greatest severity in the frontal lobes.
The GM loss is consistent with the known neuropathology of the illness, which involves cortical thinning, due primarily to a shrinkage in neuropil, and without either the neuronal loss or gliosis that characterize neurodegenerative processes (33
). The GM loss could be explained by several different neurodevelopmental mechanisms. One is pathologically extended pruning. It is well-established that normal adolescent brain development is characterized by reduction of GM and expansion of WM, processes that sculpt the brain into maturity and expand its capacity for high-level reasoning and problem solving (35
). The typical onset of schizophrenia during this time period is thought to occur as a consequence of a defect in this normal maturational process (9
), which causes abnormalities in connectivity of brain networks that, in turn, lead to the characteristic features of the illness, such as abnormalities in perception (hallucinations), inferential thinking (delusions), decreased fluency of thought and speech (alogia), or decreased emotional expression (affective blunting). If the normal process of pruning was not only defective during adolescence (overpruning) but also continued its defective actions into a later time period, such as the 20s and 30s (i.e., pathologically extended pruning), a single neurodevelopmental mechanism might explain the observation that GM is decreased at the time of onset and that it also continues to decrease during the years immediately after onset. Another mechanism that may explain the GM decreases that occur after the onset of schizophrenia is diminished neuroplasticity, an impairment in activity-dependent neuroplasticity that affects spines and synapses, leading to a shrinkage of neuropil. In this scenario, the driving force would be an impairment in the processes that regulate activity-dependent modeling of the pattern and strength of synaptic connections, with associated reduction in spines and even dendritic arbors. This would also manifest itself neuropathologically as a shrinkage in neuropil and cortical thinning without gliosis or neuronal loss. Genes that regulate neurodevelopment and neuroplasticity (e.g., BDNF, DISC1, ERBB4, NRG1
) are candidate causal agents for the GM findings (38
Another possible explanation for the brain tissue loss is that it is a consequence of some confounder, such as neuroleptic treatment. Studies of neuroleptic-treated animals and of human postmortem tissue have been used to explore this possibility (40
). While the postmortem studies have generally not shown a relationship between treatment intensity and tissue change, controlled animal studies have indicated that neuroleptic-treated animals have brain volume reductions. We have examined treatment effects in the current sample and have found that both follow-up duration and neuroleptic treatment contribute significantly and independently to the tissue volume reductions, while other possible confounders such as substance misuse do not (44
). Therefore, while neuroprogression may be partially accounted for by a medication effect, it also reflects an intrinsic and progressive disease process.
Although multiple brain regions display progressive changes in GM, the most severe loss occurs in the frontal lobes. The frontal lobes are the last brain region to mature in human beings, and this maturational process also occurs during adolescence and early adulthood. It is noteworthy that we find progressive brain change after onset in both frontal lobes and in the thalamus, because prefrontal cortex receives major input from the thalamus (45
). In schizophrenia, these two closely interconnected regions, which are already decreased in size in first-episode patients, continue to decrease further after onset.
We have also found that progressive brain changes affect WM even more pervasively than GM. This finding adds to the growing evidence that schizophrenia is a disease of abnormal structural connectivity and is both a WM and GM disorder (12
). These WM findings cannot be directly explained by aberrant neurodevelopmental processes, such as overpruning, pathologically extended pruning, or diminished neuroplasticity, because these mechanisms affect GM. In fact, during normal adolescent brain development, WM is expanding while GM is declining; this maturational process is particularly prominent in the frontal lobes (32
). One possible explanation for a decline of WM after the clinical onset of schizophrenia is that normal processes are simply diminished, e.g., WM expansion is reduced, because of an impairment in myelination and factors that affect it, such as the function of oligodendroglia and the genes that regulate myelination (e.g., NRG1, ERBB3, ERBB4
This study has a variety of limitations. We were not able to obtain scans for each subject at all time points, making the analysis of progression over time more complex and potentially less definitive. Control subjects and patients were not perfectly matched; the control subjects were slightly older and included a larger proportion of female subjects, adding further complexity to the analysis of progression. In the analysis of the pattern of change over time, the sample sizes of both patients and control subjects become progressively smaller over time, potentially reducing the power to detect changes at later time points. The surveillance period is limited to the first 12 to 15 years after onset or an age range of the 20s to early 40s; this does not permit us to determine if progression again accelerates in patients during later life and leads to more rapid later deterioration. The amount of change over time (when calculated in a linear fashion as percent change per year) is less than that reported in many studies; as indicated in and , this type of linear calculation oversimplifies the measurements; on the other hand, very few other studies have measurements on such large samples at so many time points. Finally, none of the analyses has been corrected for multiple comparisons. In general, a priori directional hypotheses had been made for the brain measure analyses, thereby lessening the need for Bonferroni corrections. The analyses examining the relationship between brain measures and clinical variables are regarded as exploratory and require replication in subsequent studies.
Figure 2 Frontal white matter (WM) volume reductions in schizophrenia patients were most pronounced early in the course of schizophrenia. Frontal WM reductions in schizophrenia patients differed significantly from healthy volunteers during the first interscan (more ...)
Our findings have multiple clinical implications. First, schizophrenia is phenotypically diverse at the neural level; progressive brain change occurs only in a subset of patients. This subset may represent a biologically distinct subgroup of schizophrenia patients that could be designated as having neuroprogressive schizophrenia and that may have a distinct pathophysiology and molecular and cell biology. Basing phenotype definition on measures of progressive brain change may offer a more powerful strategy for identifying genetic mechanisms than symptom patterns or DSM diagnoses. It is also important to note that the majority of patients are not losing brain tissue at an abnormally rapid rate, a finding that offers good news. Second, progressive brain tissue loss is correlated most closely with cognitive performance; its relationship with symptom dimensions or remission is more modest. These clinical associations provide some additional support for defining neuroprogressive schizophrenia as a biologically meaningful subtype but also suggest that symptom measures may provide a relatively weak approach for phenotype definition. Third, the losses are greatest during the early stages of the illness, and they may occur as a consequence of impaired neuroplasticity or structural or functional connectivity in the brain as a whole but particularly the frontal lobes and the thalamus. This suggests a need to continue the ongoing search for treatments that might be particularly effective during early phases and that might enhance connectivity, neuroplasticity, and cognition. Early use of cognitive rehabilitative strategies may be a promising choice, while we also continue to search for pharmaceutical treatments that will directly target neurodevelopmental mechanisms.