**4. Prenatal drug exposure and brain function**

Neural circuits that control brain function have different patterns of activity that can be measured using fMRI. fMRI provides an indirect measure of brain function by quantifying the blood oxygen level-dependent (BOLD) response, which reflects changes in blood oxygen utilization throughout the brain. When neural circuits become active, MR signals will increase by a small amount, reflecting a signal change of approximately one percent. The ability to detect a change in MR signal depends on the different magnetic properties of oxygenated vs. deoxygenated blood and that blood flow to areas of the brain that are working are very sensitive. Different types of experimental designs are used in conjunction with fMRI meth‐ ods to determine the location of brain activity. In the simplest type of experiment, patterns of brain activity are examined as a subject alternates between an experimental and control condition. The signal will increase and decrease as a function of the experimental conditions after adjusting for time. Functional neuroimaging studies produce group-averaged maps that show the level of brain activation that is associated with a specific task or in response to a specific stimulus. The group maps are then compared between conditions and/or between groups to examine the magnitude and extent of brain activation for a given response [96].

fMRI research has been used to determine if prenatal drug exposure has an impact on areas of the brain that receive more or less oxygenated blood in response to performing a cogni‐ tive task. The method has been used to demonstrate the effect of prenatal drug exposure on brain activation during a variety of cognitive behaviors. Recent work converges on three domains, the default mode network, inhibitory control, and working memory; all of which illustrate how fMRI methods can be used to better understand the impact of prenatal drug exposure on brain function. In addition, innovative functional connectivity studies have combined information from structural (MRI and DTI) with functional (fMRI) methods to understand the temporal relations between spatially distinct brain regions.

#### **4.1. Default mode network**

The default mode network (DMN) is comprised of a set of brain regions including ventral medial prefrontal cortex, posterior cingulate, inferior parietal lobe, lateral temporal cortex, dorsal medial prefrontal cortex, and the hippocampus (see **Figure 1**) [97]. This network is active when one appears to be atrest butis actually engaged in spontaneous and goal-directed mental tasks such as free-thinking, remembering, and making future plans [98]. In contrast, the network is inhibited while performing tasks with high-cognitive demand and increased task difficulty [99, 100]. Behaviorally, both prenatal cocaine and alcohol exposure are associated with early and persistent deficits in arousal regulation and attention deficits [101–105] and an increased risk for a diagnosis of attention-deficit/hyperactivity disorder [106, 107]. One interpretation of the results of these studies is that the dysregulation of arousal and attention, in part, explains other observable deficits in higher-cognitive function.

controlled carefully for other prenatal drug exposures showed that prenatal cocaine-related FAdifferences in fiber pathways including right cingulum, right arcuate fasciculus, leftinferior longitudinal fasciculus, and splenium of the corpus callosum were associated with deficits in attention and response inhibition [94]. Only one study has reported a trend for higher FA associated with prenatal methamphetamine exposure [95]. These early studies and the lack of research on the impact of prenatal tobacco and marijuana exposure on white matter integri‐ ty indicate the need for additional research to better understand the impact of prenatal drug

Neural circuits that control brain function have different patterns of activity that can be measured using fMRI. fMRI provides an indirect measure of brain function by quantifying the blood oxygen level-dependent (BOLD) response, which reflects changes in blood oxygen utilization throughout the brain. When neural circuits become active, MR signals will increase by a small amount, reflecting a signal change of approximately one percent. The ability to detect a change in MR signal depends on the different magnetic properties of oxygenated vs. deoxygenated blood and that blood flow to areas of the brain that are working are very sensitive. Different types of experimental designs are used in conjunction with fMRI meth‐ ods to determine the location of brain activity. In the simplest type of experiment, patterns of brain activity are examined as a subject alternates between an experimental and control condition. The signal will increase and decrease as a function of the experimental conditions after adjusting for time. Functional neuroimaging studies produce group-averaged maps that show the level of brain activation that is associated with a specific task or in response to a specific stimulus. The group maps are then compared between conditions and/or between groups to examine the magnitude and extent of brain activation for a given response [96].

fMRI research has been used to determine if prenatal drug exposure has an impact on areas of the brain that receive more or less oxygenated blood in response to performing a cogni‐ tive task. The method has been used to demonstrate the effect of prenatal drug exposure on brain activation during a variety of cognitive behaviors. Recent work converges on three domains, the default mode network, inhibitory control, and working memory; all of which illustrate how fMRI methods can be used to better understand the impact of prenatal drug exposure on brain function. In addition, innovative functional connectivity studies have combined information from structural (MRI and DTI) with functional (fMRI) methods to

The default mode network (DMN) is comprised of a set of brain regions including ventral medial prefrontal cortex, posterior cingulate, inferior parietal lobe, lateral temporal cortex, dorsal medial prefrontal cortex, and the hippocampus (see **Figure 1**) [97]. This network is active when one appears to be atrest butis actually engaged in spontaneous and goal-directed mental

understand the temporal relations between spatially distinct brain regions.

exposure on DTI measures.

**4.1. Default mode network**

**4. Prenatal drug exposure and brain function**

200 Recent Advances in Drug Addiction Research and Clinical Applications

**Figure 1.** Key regions associated with the default mode brain network.

Current neuroimaging research suggests, however; that the underlying impact of prenatal cocaine or alcohol exposure on arousal and attention reflects changes in function of the DMN network. Results are summarized in **Table 1(A)**. Using resting-state fMRI, a recent large-scale study of neonates with prenatal cocaine exposure or polydrug exposure showed polydrugrelated connectivity disruptions within frontal-amygdala, frontal-insula, and insula-sensori‐ motor circuits; and specific effects of prenatal cocaine exposure on the frontal-amygdala network [108]. Results showed that polydrug exposure was associated with negative connec‐

tivity within these networks. Negative connectivity is interpreted as a dysregulation in excitatory and inhibitory inputs [109–111], and in this case, a failure to inhibit the amygdala response from prefrontal cortex inputs.

A number of studies indicate that the effect of prenatal cocaine exposure on functional differences within the DMN persistinto childhood and adolescence. Adolescents with prenatal cocaine exposure show overall reductions in regional cerebral blood flow at rest with compensatory, relative increases in anterior and superior brain regions [112]. Additionally, while in the resting state, adolescents with prenatal cocaine exposure show increased functional connectivity in the DMN compared to controls [113], and less deactivation of the network in the DMN, while performing a working memory task with emotional distracters.

Furthermore, the effects of prenatal cocaine and alcohol exposure on the DMN can be dissociated. Similar to prenatal cocaine exposure, prenatal alcohol exposure is associated with less deactivation in the DMN while performing a cognitive task [114]. In contrast to prenatal cocaine exposure, prenatal alcohol exposure is associated with decreased functional connec‐ tivity within the DMN at rest [114]. These results suggests that the underlying mechanism for prenatal cocaine or alcohol exposure effects on cognitive ability are due, in part, to changes in baseline levels of arousal and dysregulation of excitatory and inhibitory control of neural resources allocated to perform cognitive tasks.

#### **4.2. Inhibitory control**

The ability to engage in voluntary, goal-directed behavior requires activation of neural circuitry that supports cognitive control mechanisms. Response inhibition is considered to be a key component of cognitive control and refers to the ability to inhibit a response that is no longer needed or inappropriate given a change in either internal or external states [115]. The go/no-go task is a cognitive paradigm that can be used in conjunction with fMRI to evaluate response inhibition [115, 116]. In the go/no-go task, participants are required to respond or withhold a response depending on whether they are presented with a "go" stimulus or a "nogo" stimulus, respectively.

The go/no-go task has been used to determine independent effects of prenatal alcohol, cocaine, marijuana, and tobacco on response inhibition, allowing for a comparison across studies. Results are summarized in **Table 1(B)**. Children with prenatal tobacco [117] or marijuana [118] exposure were more likely to commit commission errors while performing the go/no-go task, but children with prenatal alcohol or cocaine exposure showed no behavioral differences in task performance. Prenatal alcohol exposure was associated with increased brain activation in prefrontal regions and less activation in the caudate compared to controls [119]. A similar pattern is demonstrated in adolescents with prenatal alcohol exposure suggesting long-term changes in brain function associated with response inhibition [120]. In contrast, prenatal cocaine exposed children showed increased activation in inferior frontal cortex and caudate and less activation in temporal and occipital regions [121]. Prenatal marijuana was associat‐ ed with differential activation of frontal regions including and increased BOLD response in bilateral the prefrontal cortex and right premotor cortex, and a decreased response in the cerebellum [118]. Children with prenatal tobacco exposure showed increased activation in a

more diverse set of brain regions including left frontal, right occipital, bilateral temporal, and parietalregions, and less activation in the cerebellum [117]. Young adults with prenataltobacco exposure showed a similar pattern ofresults with increased activation inferiorfrontal, inferior parietal, basal ganglia, and cerebellum [122].

Results across multiple studies indicate that prenatal drug exposure leads to differential activation in frontal–striatal circuits, while performing the go/no-go task. In addition, across studies, prenatal drug-related increases in activation were reported in many brain regions, which indicates an increase in the demand for cognitive resources, while performing the response inhibition task. This pattern of results is indicative of an immature brain circuitry. Across development, the typical pattern observed in neuroimaging data is that for response inhibition, there is an increase in the magnitude of activation and a decrease in the extent of activation in frontal–striatal brain regions [123, 124]. Increased efficiency of neural process‐ ing is also associated with a peak in behavioral performance. Younger children show greater activation in similar brain regions as reported in the prenatal drug imaging studies [125, 126]. Although, the data collected in each of the studies were cross-sectional, the reported effects of prenatal drug exposure in childhood, adolescence, and adulthood indicate that the changes in brain circuitry underlying response inhibition may not be due to developmental delay, but instead due to long-term changes in the activation of the circuit.

#### **4.3. Working memory**

tivity within these networks. Negative connectivity is interpreted as a dysregulation in excitatory and inhibitory inputs [109–111], and in this case, a failure to inhibit the amygdala

A number of studies indicate that the effect of prenatal cocaine exposure on functional differences within the DMN persistinto childhood and adolescence. Adolescents with prenatal cocaine exposure show overall reductions in regional cerebral blood flow at rest with compensatory, relative increases in anterior and superior brain regions [112]. Additionally, while in the resting state, adolescents with prenatal cocaine exposure show increased functional connectivity in the DMN compared to controls [113], and less deactivation of the network in the DMN, while performing a working memory task with emotional distracters. Furthermore, the effects of prenatal cocaine and alcohol exposure on the DMN can be dissociated. Similar to prenatal cocaine exposure, prenatal alcohol exposure is associated with less deactivation in the DMN while performing a cognitive task [114]. In contrast to prenatal cocaine exposure, prenatal alcohol exposure is associated with decreased functional connec‐ tivity within the DMN at rest [114]. These results suggests that the underlying mechanism for prenatal cocaine or alcohol exposure effects on cognitive ability are due, in part, to changes in baseline levels of arousal and dysregulation of excitatory and inhibitory control of neural

The ability to engage in voluntary, goal-directed behavior requires activation of neural circuitry that supports cognitive control mechanisms. Response inhibition is considered to be a key component of cognitive control and refers to the ability to inhibit a response that is no longer needed or inappropriate given a change in either internal or external states [115]. The go/no-go task is a cognitive paradigm that can be used in conjunction with fMRI to evaluate response inhibition [115, 116]. In the go/no-go task, participants are required to respond or withhold a response depending on whether they are presented with a "go" stimulus or a "no-

The go/no-go task has been used to determine independent effects of prenatal alcohol, cocaine, marijuana, and tobacco on response inhibition, allowing for a comparison across studies. Results are summarized in **Table 1(B)**. Children with prenatal tobacco [117] or marijuana [118] exposure were more likely to commit commission errors while performing the go/no-go task, but children with prenatal alcohol or cocaine exposure showed no behavioral differences in task performance. Prenatal alcohol exposure was associated with increased brain activation in prefrontal regions and less activation in the caudate compared to controls [119]. A similar pattern is demonstrated in adolescents with prenatal alcohol exposure suggesting long-term changes in brain function associated with response inhibition [120]. In contrast, prenatal cocaine exposed children showed increased activation in inferior frontal cortex and caudate and less activation in temporal and occipital regions [121]. Prenatal marijuana was associat‐ ed with differential activation of frontal regions including and increased BOLD response in bilateral the prefrontal cortex and right premotor cortex, and a decreased response in the cerebellum [118]. Children with prenatal tobacco exposure showed increased activation in a

response from prefrontal cortex inputs.

202 Recent Advances in Drug Addiction Research and Clinical Applications

resources allocated to perform cognitive tasks.

**4.2. Inhibitory control**

go" stimulus, respectively.

Working memory refers to the cognitive ability to hold and manipulate information for a short period of time. Brain imaging studies have shown a load-dependent role for the prefrontal cortex in working memory [127, 128]. Using fMRI methods, prenatal drug exposure is associated with differential brain activation within the prefrontal cortex, while performing working memory tasks. Results are summarized in **Table 1(C)**. For example, children prenatally exposed to tobacco experience more activation in the inferior parietal regions of the cortex, whereas children not exposed showed activation in the bilateral inferior frontal region of the cortex [129]. Prenatal marijuana is also associated with patterns of increased activation associated with working memory including the inferior and middle frontal gyri [130].

fMRI methods have also been used to demonstrate specific effects of prenatal drug exposure in both the visual–spatial and verbal working memory domains. Prenatal alcohol exposure leads to increased activation ofthe frontal–parietal–cerebellar network including the left dorsal frontal and left inferior parietal cortices, and bilateral posteriortemporalregions during verbal working memory compared to controls [131]. The results showed that individuals prenatal‐ ly exposed to alcohol recruit a larger, more extensive neural network than their peers. Across three studies, prenatal alcohol exposure was also associated with differential patterns of activation, while performing spatial working memory tasks [132–134]. In contrast, offspring with prenatal methamphetamine exposure had less activation than their unexposed counter‐ parts in both the frontal and striatal regions; primarily in the left hemisphere of the brain on a spatial working memory task [135], but increased activation in bilateral temporal regions in response to performing a verbal working memory task [136].


**Table 1.** Summary of prenatal drug exposure effects on (A) default mode network, (B) working memory, and (C) inhibitory control.

The impact of prenatal alcohol exposure can be dissociated from other potential explanatory variables. When examined in relation to family history of alcohol use disorders, prenatal alcohol exposure independently predicted increased activation in left middle and superior frontal brain regions [137]. In a direct comparison of adolescents with prenatal alcohol exposure or ADHD, behavioral profiles were similar but the two groups showed differences in how cortical brain regions were recruited for spatial working memory [138]. Overall, prenatal alcohol exposure was associated with an increased effort to compensate in relation to increasing task demands compared to the ADHD group.

**(A) Default mode network**

Alcohol Increased activity in DMN

DMN at rest

DMN at rest

regions

Alcohol Increased activity in prefrontal

Tobacco Increased activity in left frontal,

and parietal regions

cerebellum

**(C) Working memory**

Cocaine Increased activity in inferior

Decreased activity in the

frontal cortex and caudate

temporal and occipital regions

Decreased activity in

Alcohol Increased activation in bilateral dorsal frontal, bilateral posterior temporal, and left inferior parietal regions

> parietal cortex as opposed to bilateral inferior frontal

Decreased activation in frontal

particularly in left hemisphere

and striatal regions,

Tobacco Activation of inferior

cortex

Methamphe tamine

inhibitory control.

**(B) Inhibitory control**

Cocaine Increased activity in

during cognitive tasks

204 Recent Advances in Drug Addiction Research and Clinical Applications

Decreased activation of

DMN during cognitive tasks

Decreased activity in the caudate

right occipital, bilateral temporal,

Increased activation of

**Drug Effects on network Behavioral effects References**

Increased risk of ADHD diagnosis

Increased risk of ADHD diagnosis

Increased effort required for

Increased effort required for

More effort required to maintain

Different mechanisms are employed to maintain working memory

response inhibition

More likely to commit commission errors

response inhibition

working memory

Decreased working memory performance

**Table 1.** Summary of prenatal drug exposure effects on (A) default mode network, (B) working memory, and (C)

Deficits in arousal regulation [101–107, 114]

Deficits in arousal regulation [101–107, 113, 114]

[117]

[121]

[131, 137, 138]

[129]

[135]

Increased effort required for response inhibition [119]

Alterations in behavioral and brain function measures of working memory extend to prena‐ tal cocaine exposure as well. The aforementioned deficits in arousalregulation associated with prenatal cocaine exposure appear to underlie brain and behavior-related working memory function. Li et al. [139] showed differential patterns of activation as a function of emotion– memory interactions. Increased demands on memory load diminished emotion-related activation in the amygdala in controls but not in the exposed group. In contrast, the exposed group failed to show an expected decrease in activation in the prefrontal cortex as memory load decreased in the presence of emotional stimuli. Results suggest that the impact of prenatal cocaine exposure on arousal regulation acts through both the dorsal cognitive and ventral emotional systems.

Overall, multiple studies demonstrate the complexities of prenatal drug-related effects on working memory. Patterns of brain activation associated with working memory are differ‐ ent by type of prenatal drug exposure, are present in the absence of behavioral differences, and show more extensive networks of activation compared to controls. Specific alterations in prefrontal cortex activation in response to working memory demand suggestthat these regions are taxed to a greater degree as a result of prenatal drug exposure. Furthermore, changes in activation remained after controlling for other explanatory variables such as intelligence. Collectively, studies demonstrate that the effect of prenatal drug exposure on brain activa‐ tion associated with working memory is less efficient and that increased levels of activation serve to compensate for any deficits in working memory function. Compensatory action, however, may not be sufficient in real-life situations characterized by increased demands on working memory function.

#### **4.4. Novel applications of imaging methods and statistical techniques**

Recently, a number of novel applications of functional neuroimaging and statistical methods have been employed to improve upon the limitations of current methods in detecting the subtle effects of prenatal drug exposure on brain function, develop connectivity maps, and aid in diagnosis. First, a variety of model-based or data-driven methods have been employed to analyze functional neuroimaging data. General linear modeling has been used most widely because it is effective, simple, and robust [140]. However, typical approaches to the statisti‐ cal analysis of fMRI data are limited in that they are not able to detect activation in heteroge‐ neous brain regions that have the potential to play diverse roles in multiple types of task performance [141]. A recent study successfully demonstrated the advantages of group-wise sparse representation of fMRI data and statistical coefficient mapping to evaluate the effect of prenatal alcohol exposure on functional activity. The advantages reported for this method included increased adaptability, more systematic in detecting diverse brain networks, and better able to identify commonalities and differences across subjects and groups [141].

fMRI data can also be analyzed to show how components of a neural system are working together when performing a specific task. The identification of associations between anatom‐ ically distinct time series is referred to as "functional connectivity" [140]. The ability to identify consistent, reproducible, and accurate regions of interest is the key to developing connectivi‐ ty maps [142]. Using a new strategy to develop cortical landmarks (dense individualized and common connectivity-based cortical landmarks, DICCOLs), Li et al. [143] used functional connectomics signatures to identify 10 brain regions with structurally disrupted landmarks that could be used to distinctly identify prenatal cocaine exposed brains from that of controls.

Finally, a novel application of machine learning has been used to test whether brain images can be used to correctly identify prenatal cocaine-exposed young adults from socioeconomi‐ cally matched controls [144]. Regional features were extracted from both structural and functional MR images, and the power of each to discriminate between prenatal cocaine exposed and control brains was accomplished through machine learning methods. The method accurately identified 91.8% of prenatally cocaine-exposed brains. The use of both structural and functional images was critical to improving the accuracy of the classification system compared to either type of image alone.
