Taken together, these results suggest that prenatal cocaine exposure may lead to regionally specific patterns of morphological changes in the striatum and subtle volumetric differences in certain frontal cortical regions. The most significant finding in our analyses of caudate morphology was an association between levels of prenatal cocaine exposure and surface contraction in the ventromedial and dorsolateral caudate. The ventromedial caudate is part of the lateral orbitofrontal striatal loop, which is involved in the regulation of emotion and social behavior
. Interestingly, we also reported regional volume changes in the left lateral orbitofrontal cortex in the PCE group compared to controls. The dorsolateral caudate, on the other hand, is part of the executive loop associated with higher-order cognitive functions
, and we also found local volumetric differences in the frontal poles, which play a role in spatial working memory, response inhibition
, and the evaluation of self-generated
 and goal-directed
 decisions. Therefore, the findings presented here may represent some of the neural correlates of the difficulties in emotional regulation
[5, 6] and impairments in attention, response inhibition
, and visuospatial working memory
 that have been reported in children with prenatal cocaine exposure.
The putamen is part of the fronto-striatal loop involved in motor control
. Most premotor area projections are directed to the medial putamen, and most supplementary motor area projections terminate in the posterior putamen
, and in both subregions we found that greater prenatal cocaine exposure was associated with a contraction of striatal surfaces. In addition, we observed significant reductions in regional volumes in the bilateral caudal middle frontal cortices in the PCE group compared to controls, and the caudal part of the middle frontal gyrus corresponds to premotor brain areas. Therefore, it is possible that these findings may be related to the deficits in fine motor coordination that have been reported in this population
While the direction of changes in striatal surface structure were not predicted a priori, it should be noted that prenatal cocaine exposure was associated with surface contraction in some subregions, and expansion in others. Though the biological mechanisms contributing to these findings remain unclear, the bidirectional nature of regional effects may explain why overall differences in striatal volume were not detected in this study between exposed and control participants in either the surface-based or the volumetric analyses. Similarly, while we did not have specific predictions about the direction of changes in regional frontal cortical volumes, prenatal cocaine exposure was shown to be associated with volume reductions in some frontal subregions, and increased volumes in others. The reasons for this discrepancy remain unclear, but the localized and bidirectional nature of frontal cortical effects may explain in part why certain neuroimaging studies found significant structural differences in the frontal lobes, while others reported negative results.
Consistent with our predictions, we observed very narrowly localized correlations between measures of executive functioning and regional deformation of the dorsal caudate surface. However, the specific areas where this association was significant did not correspond exactly to the subregions of the dorsal caudate where higher levels of prenatal exposure were correlated with greater dysmorphology. The specific areas of the dorsal caudate showing correlations with measures of executive functioning also differed by task: they were more superior and medial for the Stroop test than for part A of the Trail Making test.
As predicted, we also observed a marginally significant correlation between a measure of visuomotor performance and regional deformation of the putamen surface, which was significant in the medial and posterior putamen, where most premotor and supplementary motor area projections terminate
, suggesting a possible association between neurological and behavioral abnormalities.
However, while these results suggest that prenatal exposure to cocaine may affect the morphology of the striatum and regional frontal lobe volumes, it is important to keep in mind that the effect sizes were small for all of the results reported here, and that the surface-deformation and volumetric maps were not corrected for multiple spatially correlated comparisons. Nevertheless, the fact that we reported subtle changes consistent with findings from the animal literature and with our a-priori hypotheses, suggests that these differences may be due in part to the specific effects of prenatal cocaine exposure.
We found no association between PCE and neurological test scores, which suggests that brain structure may be a more sensitive biomarker to levels of prenatal cocaine exposure than neuropsychological test performance. This is consistent with findings from the animal literature, suggesting that cognitive tests may be more sensitive to the pattern of maternal consumption than to the amount of cocaine intake, even in the presence of neurobiological alterations
. Despite showing evidence for abnormal neuronal migration and cortical lamination as well as neurochemical differences, rhesus monkeys with both ‘high doses’ and ‘low doses’ of PCE do not significantly differ from controls in cognitive performance, whereas monkeys in the ‘escalating dose’ group show impairments
It should be noted that we did not have data about participants’ use of other drugs, thus we cannot exclude the possibility that postnatal exposure to other substances of abuse may have affected the findings. Another important limitation is that neuroimaging data was available for only 12 adolescent controls. Though we cannot rule out the eventuality that a larger control group would have allowed for the detection of slightly more robust group differences, in the context of the existing literature, these findings support the notion that the effects of PCE on brain structure may be quite subtle. An alternative explanation to findings of modest effects in the current study may be that the marginally significant alterations observed simply reflect relatively minor consequences of PCE on brain development. Some publications suggested that cocaine may be a relatively weak teratogen with few observable neurological or behavioral consequences in humans
, unlike other common substances of abuse during pregnancy, which have been more convincingly linked to psychopathology risk, such as alcohol
 and nicotine
The small effect sizes observed here as well as regional differences in the direction of effects may help explain why prior investigations of the consequences of prenatal cocaine exposure on brain structure have yielded somewhat conflicting results. Although animal studies of PCE, particularly studies of non-human primates, clearly demonstrate the potential of prenatal cocaine exposure to interfere with brain development at a cellular and biochemical level in various brain regions
[24, 25, 47, 50–52], they also suggest that the types and severity of PCE effects largely depend on the route, dose, gestational period, and pattern of consumption
. These could be additional contributing factors to the discrepancies in the existing human neuroimaging literature, and to the small effect sizes reported here.
Thus, it is important that future neuroimaging studies of prenatal cocaine exposure with larger samples collect information about adolescent participants’ use of other substances, specific patterns and timing of maternal cocaine consumption, and aim to integrate observations from different brain imaging modalities. This will help determine how structural, metabolic, and functional brain abnormalities resulting from PCE relate to real-life difficulties these children face outside the scanner, as subtle neurological changes may very well be associated with important behavioral, cognitive, or emotional impairments.
Although several promising psychosocial prevention strategies for pregnant women addicted to cocaine have been identified
, effective remediation strategies and treatments for prenatally exposed children remain to be developed. The improvement of such strategies will require that we gain a better understanding of the specific, localized and perhaps subtle neurological abnormalities resulting from prenatal cocaine exposure in order to facilitate the translation of research findings to clinical practice.