The brainstem is a central structure that connects the brain cerebrum to the spinal cord and cerebellum. The brainstem contains nuclei and fiber pathways that synthesize and transfer specific neurotransmitters and neuromodulators to the peripheral receptors and effectors for the regulation of such basic functions as arousal, motor function, memory, reward, nociception, and autonomic control. Along a dorsoventral axis, the brainstem consists of three subdivisions with distinctive parts: the dorsal side of the midbrain known as the tectum, including the mid-dorsal part of the aqueduct, a central region of nuclei and fibers known as tegmentum beneath the ventricular system, and a massive fiber communication system connecting to the spinal cord and the cerebellum in the ventral part. Nuclei and fiber pathways in the tectum relay pain signals from peripheral stimuli via the spinal cord to the cerebral sensory cortices that result in the sensation of pain (nociception). They also transmit anti-nociceptive signals down to the spinal cord which induces endogenous opioid-based analgesia (pain regulation). Many pain syndromes can be caused by damage to or dysregulation of the brain, brainstem, spinal cord and fibers connecting them. Invasive (surgical) inspection of such damage is often carrying considerable risk which is why non-invasive imaging techniques are a very attractive alternative.

Conventional brain structural MRI studies have assessed how gray matter volume and white matter integrity are associated with acute pain processing or the severity in chronic pain. Diffusion tensor imaging (DTI) is a specific structural MRI sequence that allows for non-invasive measurement of altered microstructural integrity in particularly white matter regions and tracts in human brain. DTI measures such as fractional anisotropy (FA) explain the strength of the orientational organization of microstructures, radial (RD), axial (AD) and mean (MD) diffusivities are likely to be sensitive to myelin/axon as well as non-specific physiopathology states. In neurological conditions, it is generally believed that reduced FA and increased MD, RD or AD are attributed to demyelination, loss of axons (1, 2), or an interrupted connection (3), which might represent potential damage of the structure in the testing anatomy. Several DTI studies (see Table 1), using non-hypothesis-driven whole brain voxel-wise analysis, found significant microstructural damage in many cerebral regions/fiber connections (some including brainstem fiber tracts) in painful conditions, such as temporomandibular disorders (4), episodic cluster headache (5) or traumatic brain injury-induced chronic headache (6), ankle muscle proprioception in low back pain (12), chronic irritable bowel syndrome (7), chronic pelvic pain (8), cervical spondylosis-induced pain (9), chronic musculoskeletal pain (10), and analgesia in response to pain stimuli (11). These studies have suggested that microstructural damage in the brainstem fiber connections may be associated with pain regulation or pain sensation. Figure 1 indicates the brain tracts that were found to be significantly affected by pain conditions from all published clinical studies reviewed in this article. The majority of these findings were reported in large fiber bundles and brainstem-to-cerebral cortical connections (16).

To date, the human brainstem remains one of the most challenging regions to explore with DTI due to the difficulty of accurately delineating small nuclei and complex fiber connections in this area. It is thus necessary to create an atlas of the brainstem small nuclei/fiber tracts in standard neuroimaging space (e.g., in MNI space) (17, 18), and perform imaging analysis by co-registering individual DTI to the MNI space (or vice-versa).

This mini-review is aimed to provide a brief overview on recent advances in brainstem tractography based on DTI, particularly focusing on the technical improvement of automated tractographic approaches. We also discuss the neuroanatomical relevance of brainstem tractography to pain, practical applications and potential implications of brainstem tractography for clinical studies of pain. We hope this article will improve the understanding of the basic principles, help with result interpretation, and increase the appreciation of technical advantages/limitations in DTI tractography. A critical understanding of this technique will promote its use and foster its application in the clinical setting for surgical interventions (19, 20) treatment and evaluation.

Many experimental and preclinical studies have investigated brainstem small nuclei and circuits that are involved in pain (79), including the periaqueductal gray matter (PAG), rostral ventromedial medulla (RVM), locus coeruleus (LC), dorsal horn (DH) of the spinal cord, and a PAG-RVM-DH circuitry/axis that interconnects them. The PAG sends direct and indirect projection to RVM and DH (80). Part of PAG is involved in an endogenous analgesia, as direct stimulation of the PAG can suppress pain intensity in individuals with various chronic pain conditions (81–83). Numerous studies have suggested a potential mechanism of descending pain inhibition and its pathways through the PAG-RVM-DH axis (84–88). In contrast to the role of reducing pain, some experimental studies have suggested that excitation of the PAG-RVM-DH system can lead to a facilitating effect through the DH (89, 90), SpV (91), and RVM (92) resulting in hyperalgesia.

To extend our understanding of the PAG-RVM-DH pathways gained through experimental studies in animals, human brain functional MRI (fMRI) studies have been promoted to investigate brainstem functional activation patterns associated with pain sensitivity and pain modulation. An extensive number of fMRI studies have reported the brainstem being involved in descending analgesic responses in various pain conditions. For example, a resting state-fMRI study (93) reported patients with chronic orofacial pain show increased functional connectivity between the RVM, LC and PAG, as well as supratentorial regions (e.g., hippocampus, nucleus accumbens, and anterior cingulate cortex), possibly reflecting “top–down” engagement of the circuitry alongside altered reward processing in pain conditions. Other studies found that higher functional connectivity between PAG and RVM (or LC) is associated with higher pain scores in fibromyalgia (94), as well as in chronic neuropathic orofacial pain (93) and diabetic polyneuropathy-induced pain (95). Several studies (96, 97) investigated healthy volunteers during heat stimulation and distraction (e.g., participants were asked to perform a color-word Stroop distractor task while receiving thermal stimuli) and found that distraction was associated with a significant reduction of pain intensity with increased functional activation of the cingulate, hippocampus, thalamus, PAG, and brainstem regions, exerting a “top-down” influence on pain modulation during distraction. In addition to the PAG-RVM-DH pathways, several fMRI studies also revealed that greater functional connectivity of the medial prefrontal cortices to the amygdala and nucleus accumbens are associated with severer acute (98) and chronic pain (99–101).

The neuroanatomical and neuropsychological studies of pain (102) suggest a complex network comprised of ascending and descending pathways in the brain. The descending pathway, also known as the “top down” pathway via the PAG-RVM-DH axis, has been demonstrated to be an endogenous analgesic system that is activated in response to pain stimuli. The ascending pathways transmit nociceptive signals from peripheral nerves to the sensory cortex via the dorsal horn of the spinal cord, brainstem, and thalamus (103). While these complex neurotransmitter pathways can be represented through tractography, our understanding of the anatomical connectivity and its implication in pain processing or pain control is still an emerging field of exploration.

Recent improvements in diffusion tensor tractography allow presentation of major brainstem pathways in humans. Given its role in pain modulation several studies performed diffusion tractography of the PAG manually (104) or a PAG site used in deep brain stimulation (DBS) (105–107) for treatment of chronic pain to identify fiber tracts that might be associated with pain. Tracts that are consistently presented in these studies include: (1) PAG to the thalamus, which overlaps with the superior cerebellar peduncle (SCP) and the spinothalamic tract (STT); (2) PAG to the medial prefrontal cortex through ventral tegmental area (VTA) and nucleus accumbens, which overlaps with the medial forebrain tract (MFT); (3) PAG to the hypothalamus and RVM, which overlaps with the dorsal longitudinal fasciculus (DLF) and partially the medial longitudinal fasciculus (MLF). Further anatomical connections with the PAG beyond the brainstem area include amygdala, anterior cingulate cortex, ventromedial prefrontal cortex (which overlaps with MFT), ventral posterior thalamus and primary somatosensory cortex (which overlaps with STT). A schematic map of the tracts that possibly relate to pain displayed in Figure 2.

On the basis of existing automated brainstem DTI tractographic atlases, Figure 3 presents an illustration of the fiber tracts that are highly relevant to pain and their relationship with key brainstem nuclei (e.g., PAG, LC and RVM). The anatomy of these brainstem tracts can be confirmed with known neuroanatomical books (108–111).

The DLF connects hypothalamus, PAG and spinal cord via the ventral side of the aqueduct and fourth ventricle. The DLF anatomy is consistent with the inferomedial branch of the medial forebrain bundle that has been described in several DBS studies (112), which consists of fiber that connects through upper pons, retrorubral area, PAG, VTA in the midbrain and ends in the lateral hypothalamus. Using atlas-based automated brainstem tractography (76), our recent study reported that FA decrease and MD/RD/AD increase in the DLF are associated with increased pain intensity in a group of patients with fibromyalgia-like, chronic pain (13). The DLF appears to have the largest anatomical overlap with the described descending analgesic pathway which in turn suggests that damage of the DLF may result in impaired endogenous pain regulation.

The SCP connects the dentate nucleus with the thalamus, red nucleus, vestibular nuclei, and the reticular formation. It functions mainly as a network for motor coordination and postural control. The SCP at the midbrain level includes non-decussated and decussated branches that partly overlap with PAG and VTA (113). One study (12) using an automated atlas-based approach, reported FA decrease and MD increase of the SCP to be significantly associated with greater sensation of low back pain in response to ankle muscle vibration in healthy volunteers.

The MFT connects between deep cerebellar nuclei and anterior frontal regions. It traverses partially through the VTA, PAG, inferomedial thalamus, also passes through the anterior limb of internal capsule (ALIC) along with anterior thalamic radiation (ATR). Similar MFT connection, which is described as a superolateral branch of the medial forebrain bundle in a previous review article, is considered to be a seeking/pleasure pathway (112). DBS treatment on this pathway showed anti-depressive effects and modified the emotional pain state, suggesting that the MFT may have therapeutical implication for affective pain. Two studies investigated MFT (or fibers with similar anatomy) changes in response to pain in healthy subjects. One using a semi-automated approach (which manually delineates tractographic landmarks from the FreeSurfer parcellation atlas), reported that FA reduction and RD increase in the MFT are associated with higher pain sensitivity after heating stimuli (14). Another study, using a voxel-wise statistical approach with a FA-based tract skeleton, showed reduced FA in several tracts (connecting dorsolateral prefrontal cortex, rostral anterior cingulate cortex with the PAG) was associated with better placebo analgesic responses (11), which supports the notion that lower FA in the MFT is associated with decreased sensitivity to painful stimuli.

The STT is known as a major sensory pathway that is highly trackable and reproducible by either manual or automated approaches. The STT consists of a spinothalamocortical pathway (pain and temperature sensation) and a medial lemniscothalamocortical pathway (conscious proprioception) (114). Both pathways are considered the main ascending sensory pathways of the body that travel rostrally in close proximity within the brainstem to the thalamus and to the somatosensory cortex (115). Several papers showed central post-stroke pain (those without signs of peripheral neuropathy) was related to injury of the STT detected by loss of the STT volume (116), thickness (117), or damaged microstructural integrity including FA decrease and MD increase (15).

Taken together, these studies applying diffusion tensor tractography in pain research demonstrated that some dorsal brainstem fibers may be involved in pain processing and modulation on the basis of these fiber tracts connecting the cerebral cortices through the PAG to the dorsal spinal cord. These studies consistently observed a significant correlation between a reduced microstructural integrity of the brainstem tracts connecting to PAG and increased pain intensity/sensitivity. These findings suggest a structural connectivity-related endogenous pain regulation. Specifically, substantial impairment (e.g., impaired axon or myeline membrane) of the brainstem fiber connection results in a dysfunction of the descending inhibition or modulation. Although the clinical application of diffusion tensor tractography in pain is still infrequent, the above evidence suggests that its inclusion in future studies may hold a lot of promise.

Current pain research applying brainstem diffusion tensor tractography has been limited in several ways: (1) Most of the brainstem tractography atlases have been built based on healthy young adults with sophisticated diffusion protocols that would last longer than 50 min. A super-high resolution (e.g., 200 Mm isotropic voxel) using higher magnetic field strengths such as 7T scanner could further improve tractographic atlas building and analyses in microstructural level but 7T scanners are typically not available in hospitals as they provide limited clinical relevance. This is further complicated by the fact that atlases build on 7T data will not directly translate into lower field scans like the commonly used 3T clinical scanners. Conversely, clinical DTI protocols routinely tend to be no longer than 10 min with lower resolution (≤2 mm³) and fewer diffusion directions (≤80). Therefore, in clinical practice, small pathways may not be identifiable and mis-registration may often occur in patients with conditions affecting brain morphology including elderly patients. Further technical developments should be focused on improvement of quality and feasibility of clinical scans, as well as correcting errors due to low resolution, noise, artifacts, distortion and crossing-fibers during post-processing. (2) Although a number of existing diffusion tensor tractographic studies have consistently attributed a role to the structural connectivity of brainstem tracts in descending analgesic pathways, less is known whether these tracts might also be associated with ascending pain facilitation effects. It is technically even difficult because the diffusion tensor tractography expresses a bidirectional structure that cannot separate the afferent or efferent fiber inputs, thus cannot distinguish the ascending or descending pathways, respectively. (3) Generally, alterations of quantitative DTI variables explain non-specific biological features and therefore only permit indirect interpretations of the underlying pathophysiology. The important role of brainstem fibers was discussed here together with supporting evidence. Additional pathophysiologic validation is needed to further understand the role of brainstem fibers in pain and other conditions.

As a non-invasive, clinically feasible MRI sequence, DTI is a sensitive imaging method in detecting problems of structural integrity of human brain neural pathways. Recent developments in diffusion tensor tractography offer visualization and quantitative analysis of the status of small brainstem fiber tracts, in which several known pain-related neural pathways are included. A better understanding of the relationship between brainstem structural connection and chronic pain and/or pain modulation will be helpful for revealing the neurobiological basis and regulation mechanisms underlying pain.

YZ and AF contributed conception and design of the review. YZ wrote the first draft of the manuscript. AF reviewed the first draft. All authors contributed to manuscript revision, read, and approved the submitted version.

This work was funded by the War Related Illness and Injury Study Center (WRIISC) of the Palo Alto Veterans Affairs Health Care System, Department of Veterans Affairs Office of Academic Affiliations and VA Clinical Science Research and Development (CSR&D) grant entitled the role of the brain stem in GWVI pathology (1 I01 CX002182-01).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.