Introduction to Multiple Sclerosis and MRI

Myelination and its role in neurological function

The central nervous system (CNS) employs electrical and chemical signals to carry out its functions. Myelin is a fatty tissue that forms a sheath around axons, the latter of which transmit electrical signals from the cell bodies out to a more distant terminus. The lipids in myelin give the tissue the properties of an electrical insulator, which ensures that charge is not dissipated over the axon length. The thicker the myelin sheath, the better insulated the axon is and the faster electrical impulses can be transmitted. Myelin is created and supported by oligodendrocytes; therefore, if either myelin or oligodendrocytes are damaged, the result is impaired electrical conduction along the axon and impaired neural communication and brain function.

What is multiple sclerosis?

Multiple sclerosis (MS) is a chronic inflammatory disease of the CNS, including brain, spinal cord, and optic nerves, that attacks myelin. Damaged myelin is being replaced by scar tissue (sclerosis) that gives the disease its name. Demyelination disrupts electrical conduction along the axon and can ultimately result in the complete transection and destruction of the nerve fiber (axon) itself. People with MS experience a range of sensory, motor, and cognitive impairments as a result.

The average age at which MS is diagnosed typically ranges from 20 to 45. An MS diagnosis requires the dissemination of symptoms across time and space. The course of MS can be very unpredictable: most patients initially present with reversible neurological deficits. From this starting point, patients segregate in terms of the progression of their symptoms. In the most common form, called relapsing-remitting MS (RRMS), patients experience flare-ups in their symptoms that remit but get progressively worse over the course of subsequent episodes. This is in contrast to the disease course known as primary progressive MS (PPMS). In these patients there are no periods of relapse or remission, but a steady worsening of symptoms and signs from onset. The cause of MS is unknown but given the role of inflammation and autoimmunity in the disease process, researchers hypothesize that MS occurs when some environmental agent (e.g., infection) interacts with a genetic makeup predisposed to immune dysfunction.

Basics of MRI

Magnetic resonance imaging (MRI) has become an invaluable tool in the diagnosis and monitoring of patients with multiple sclerosis. MRI takes advantage of the fact that brain tissue contains a large quantity of water. When a person lies in the powerful magnetic field of the MRI machine, a majority of the positively-charged protons of the water molecules align with the magnetic field. Next, a radio frequency current—called the RF pulse—is turned on and produces a varying electromagnetic field. At the resonance frequency, the RF pulse can be absorbed and knocks the spin of protons off the main axis of the magnetic field. Directly after the RF pulse, the maximum number of protons are spinning in alignment. After this, the proton spins dephase and the majority of protons go back to spinning in the direction of the static magnetic field. During this relaxation, a radio frequency signal is generated that can be picked up by receiver coils. The MRI scans we can see are reconstructed via a mathematic transformation that extracts the distribution of protons from the radio frequency signal. What gives the MRI its contrast is the fact that protons in different part of the brain behave differently within the magnetic field, and protons return to their equilibrium spin state at different rates. A variety of RF pulse sequences are used in MRI scanners to alter image contrast.

There are several important techniques involved in imaging of MS:

T1-weighted imaging: this scan shows "black holes" or lesions that are dark areas on the MRI scan, also known as hypo-intensities. Acute T1 hypo-intense lesions can be transient areas of edema that later disappear on subsequent scans. Chronic T1 hypo-intense lesions are areas of permanent myelin loss and axonal damage.

T2-weighted imaging: this scan shows bright spots, or hyper-intensities. The total number of T2 lesions and/or the T2 lesion volume can be used as a proxy for disease burden.

Gadolinium enhancement: Gadolinium is a contrast agent that is administered prior to MRI scan. The presence of gadolinium enhancement indicates active lesions, meaning that there is a breakdown of the blood brain barrier and inflammation is present.

Magnetization transfer ratio: Magnetization transfer ratio is an MRI sequence that distinguishes between the macromolecular associated water (such as that found in myelin) versus the free water pool based on the fact that the free water protons have higher spin frequencies and fewer interactions with the environment that contribute to dephasing. Therefore, free water protons can be distinguished for bound, macromolecular pools of water by their increased T2 relaxation time.

The Multiple Sclerosis Lesion

Speakers:
Istvan Pirko, Mayo Clinic
Daniel Reich, National Institute of Neurological Disorders and Stroke, NIH
Klaus Schmierer, Barts and the London School of Medicine & Dentistry

Highlights

• High resolution serial in vivo MRI reveals four distinct groups of lesion dynamics in the TMEV-induced mouse model of MS.
• The presence of a central vein in white matter lesions is a promising diagnostic tool for MS, and the patterns of gadolinium enhancement at these central veins can provide specific hypotheses for how perivascular lesions form.
• Cortical white matter lesions are underreported in MRI scans compared to histology. Use of high-field MRI imaging on post-mortem brain tissue in combination with myelin histology could inform methods of in vivo detection of cortical white matter lesions.

Tracking the MS lesion

When lesions appear in the MRI scans of MS patients they do not remain stable. Lesions can expand and retract, enhance and not enhance. The subtle changes in lesion dynamics and other features are being studied in greater detail than ever before due to the advent of higher-field MRI. By tracking lesions across time, either in animal models or human patients, researchers are interrogating the link between changes in MRI signal and lesion appearance and the link between lesion characteristics and clinical outcomes.

Istvan Pirko of the Mayo Clinic discussed the use of mouse models of multiple sclerosis to test potential clinical methodologies and therapies. His lab uses a viral infection model of MS that induces a biphasic disease in susceptible mouse strains. Upon infection, the Theiler murine encephalitis virus (TMEV) replicates in neurons in the hippocampus, striatum, cortex, and spinal cord and is capable of converting to a chronic demyelinating disease in susceptible strains. With the onset of chronic demyelinating disease epitope spreading to myelin epitopes also occurs. Upon infection with a virus like TMEV, the body mounts a viral-specific T cell-mediated immune response that causes inflammation at the site of infection. Inflammation can damage host tissues like myelin, causing local spread of broken down tissue particles. The theory of epitope spreading is that an acute viral immune response can switch to a chronic autoimmune response when T cells recognize these damaged host particles as epitopes and initiate an attack on the host tissue. Pirko's lab used MRI to track lesion load and dynamics in the autoimmunity-prone interferon-γ receptor null mouse line (G129).

Pirko saw no T2 lesions resolve in the G129 mice; instead, T2 lesion dynamics could be sorted into four groups: expanding, fluctuating, expanding-retracting, and stable. At time of death the expanding lesions accounted for the majority of total lesion volume. All fluctuating lesions showed gadolinium enhancement—indicative of blood brain barrier breakdown—when they appeared. However, after approximately 40 days enhancement was virtually unseen despite the ongoing appearance of new lesions.

The immuno-normal C57Bl/6 strain displays lesion dynamics that are distinct from the G129 line; 7 days post-TMEV infection they rapidly develop T1 black holes, regions of T1 hypointensity that indicate that axons and neurons themselves have been destroyed. Black holes weren't seen in mice that lacked an adaptive immune system, suggesting that infiltrating immune cells promote damage. T1 black holes were reduced to a greater extent when CD8+ immune cells were knocked out compared to CD4+ cells. Likewise, T1 black holes were reduced in mice null for perforin, the effector molecule of CD8+ T cells. When Pirko treated mice with TMEV viral capsid peptide prior to infection, it inhibited cytotoxic T lymphocyte (CTL) antiviral response and protected animals from paralytic disease. The researchers were thus surprised that injection of the same peptide during ongoing antiviral CTL response had dramatically adverse effects, resulting in the development of profound hemorraghic demyelination leading to mortality within two days. Ongoing work is being done to investigate differences in immune response prior to infection, in the acute phase, and chronic phase of CTL response.

Charcot revisited

One can find in the hand-drawn illustrations of Charcot, the physician who first described multiple sclerosis, the intimate association of small capillaries within sclerotic patches of white matter. Today, Daniel Reich from the National Institute of Neurological Disorders and Stroke at the NIH, and other researchers, are using high resolution MRI to study the relationship between blood vessels and MS lesions across time in patients. Studies looking at serial scans over time have reported changes in MR measures in normal-appearing white matter where focal-enhancing lesions form that are related to increased diffusion and perfusion. Postmortem data has established that acute inflammation occurs within gadolinium-enhancing lesions and that small veins are often at the center of these lesions.

Studies from the 19th century identified blood vessels in MS lesions. This illustration by JM Charcot depicts the phenomenon. (Image courtesy of Daniel Reich)

Lending further support to Charcot's original observation, Reich's group sees that small veins radiate out from the lateral ventricles and that lesions form around them. Furthermore, the presence of central veins in lesions may prove to be a useful diagnostic criterion for MS; a group in Nottingham found that patients who had veins within more than 40% of their lesions had MS. First collected at 7T, Reich has developed the FLAIR* sequence to collect small vein data in the clinical 3T scanner. The FLAIR* sequence combines FLAIR, a workhorse for finding lesions in MS, with a T2* sequence that is optimized to quickly scan and highlight veins during gadolinium enhancement. Reich measured vein diameter from inside and outside lesions within the MS brain and compared them to healthy controls. He observed significant decreases in the lumen diameter of veins found inside lesions compared to healthy control veins. However, veins outside of lesions in the MS brains appeared enlarged.

Reich's group performed serial and rapid scans immediately following gadolinium injection and found lesions exhibited two different patterns of enhancement, either a centrifugal (spread from inside the lesion, outward) or a centripetal (spread from outside the lesion, inward) pattern of enhancement. Reich hypothesized that this pattern depends on where contrast agent is sensing blood brain barrier breakdown. For example, breakdown at the central vein would result in a centrifugal enhancement. Reich's working model is that small veins become inflamed, leading to the formation of a perivascular inflammatory cuff. As soluble factors expand out from the cuff into tissue, they make other small vessels leaky. Meanwhile, repair occurs at the central vein, sealing up the blood brain barrier and causing a shift in the pattern of enhancement from centrifugal to centripetal.

How to interpret white matter lesions

Focal white matter lesions, like the perivenular ones Reich studies, are the most obvious feature of MS pathology. While clinicians can consistently detect them, it is still not clear what white matter lesions mean in terms of their cause and clinical significance. Klaus Schmierer of the Blizard Institute at Barts and the London School of Medicine and Dentistry is performing high field MRI of lesions in post-mortem tissue followed by histology to understand specific features of lesions and mechanisms underlying their genesis.

Schmierer focused on normal-appearing white matter post-mortem slices to determine the concentration of neurofilaments—the principal cytoskeletal component of axons. Specifically, he measured levels of hyperphosphorylated neurofilament heavy chains (NfH) because phosphorylation of NfH reduces the rate of axonal transport and is believed to indicate that an axon is stressed or unhealthy. Schmierer wanted to determine whether the NfH phosphorylation levels correlated with the presence of white matter lesions and if they might even precede their appearance. Levels of hyper-phosphorylated NfH were found to be higher in acute white matter lesions than in chronic lesions, and in normal appearing white matter they correlated with increased T1 and reduced magnetization transfer ratio—together indicating an increase in the free proton pool. This makes sense considering that protons can no longer bind to NfH sites that are hyper-phosphorylated. Furthermore, a decrease in magnetization transfer ratio is believed to reflect a loss of myelin content due to a depletion of the pool of protons that are associated with its macromolecular structure. Therefore, Schmierer and colleagues' finding suggests that NfH hyperphosphorylation reflects changes in axonal structure that precede and may contribute to white matter lesion formation.

Schmierer next asked if similar measurements of mean water fraction could detect remyelination events. He presented 9.4T data that showed a half-moon pattern of darker intensity covering part of a lesion that appears to indicate ongoing remyelination. While this work is ongoing, other studies suggest that there is a good correlation between measures of myelin water fraction and histological staining for myelin. However there is discrepancy in the abundance of focal demyelinating lesions in cortex versus white matter when comparing 1.5T MRI data to postmortem histology. In 1.5T MRI the ratio of cortical:white matter lesions is anywhere from 1:4 to 1:12 depending on acquisition sequences used, while it approximately the inverse (4:1) according to histology. Schmierer outlined static and dynamic sources for the difficulty in detecting cortical lesions by MRI: static issues include the small size, low myelin content compared to white matter, and low contrast when comparing to surrounding cortex; dynamic issues include the frequent occurrence of remyelination in cortical matter, inflammation, and little persistent blood brain barrier disruption in the cortex.

In cortex his group found an association between T2 and myelin content and T1 and neuronal density at 9.4T MR of cortical grey matter lesions. Now they are taking what was learned from scanning post-mortem tissue blocks back into the clinic to image hemispheres in patients. Even at 7T, however only ~1/3 of cortical lesions were reported, stressing the need for further improvements in acquisition and post-processing techniques.

Imaging Multiple Sclerosis After and Around the Lesion (Part II)

Speakers:
Daniel Pelletier, Yale University School of Medicine
Elizabeth Fisher, Cleveland Clinic
Matilde Inglese, Mount Sinai Medical Center

Highlights

• Data suggest that white matter lesions preferably form in fibers that project from the thalamus, a brain region that shows early atrophy in preclinical MS patients.
• T2 lesions moderately correlate with whole brain atrophy, but data suggest it's not the whole story.
• Spectroscopy data from brains of MS patients reveal a change in the oxidative status of the brain that is likely driven by a change in both energy demand and perfusion.

Drawing connections across time and space to uncover features of MS pathology

White matter (WM) lesions do not occur in isolation in the brain—they are one pathological feature of MS that happens to be readily detected by MRI. Multiple disease processes contribute to the formation of WM lesions, and these same processes may also drive changes in features such as grey matter volume and brain metabolism. To assess the relationships between white matter lesions and other features of MS pathology, researchers are testing correlations among different MRI markers across time and space in the brains of MS patients.

As Sean Deoni spoke about MS as a disease of brain disconnectivity, Daniel Pelletier of Yale University School of Medicine is interested in the downstream effect of white matter lesions for connected regions of grey matter. Levels of N-acetyl-aspartate (NAA) are decreased in chronic lesions, indicating that axonal integrity is threatened. Furthermore, studies have reported that white matter lesion volume correlates with deep grey matter volume loss. In a study of pediatric cases of MS, the volume of the thalamus appeared to be especially correlated with white matter lesion volume. These findings beg the question of what happens to neurons that have their axons severely damaged or transected—is there retrograde degeneration of neurons that drives grey matter atrophy?

Pelletier's group looked at CIS patients (pre-MS diagnostic patients) to see if they had brain atrophy compared to control populations. They performed regional voxel-based morphology analysis, a technique in which each pixel possesses a volume and can be registered to a spatial location on a brain atlas. Pelletier found that at the time of CIS presentation thalamic volume was lower than normal on both sides in MS patients compared to controls. These data suggest that the thalamus is a sensitive location for detecting early grey matter loss. Next, Pelletier and colleagues used DTI fiber tracking to ask if white matter tracts that connect the thalamus to other brain regions is affected in CIS patients. They delineated thalamocortical projections, creating a template into which all patient lesion and diffusion data could be registered into the same coordinate system, and found that while thalamocortical projections account for 28% of the total white matter volume, 78% of the lesion volume was found there. This identified a 10:1 bias for white matter lesions to occur within thalamocortical projections.

Furthermore, Pelletier found that they could explain 66% of the variance of thalamic atrophy with three key features: thalamocortical lesion volume, DTI changes in thalamocortical white matter lesions, and DTI changes to the normal-appearing white matter (NAWM) that connects lesions and thalami. The presence of abnormal NAWM outside of the thalamocortical pathway did not contribute to the prediction of thalamic volume. Pelletier highlighted findings that non-lesional neuronal volume is reduced in thalamic nuclei, supporting the idea that neuronal loss could be a retrograde event following axonal transection in white matter tracts projecting from the thalamus. Pelletier concluded that the thalamus could serve as a sensitive brain region in which to study neuronal degeneration in MS.

Pelletier and colleagues used diffusion tensor imaging fiber tracking to delineate thalamocortical projections and created a template into which all CIS patient lesions and diffusion data were registered. This slide shows two fiber tracts projecting from the thalamus (in green) to the cortex. Note that the tracts are seeded through white matter lesions (shown in blue). Mean diffusivity is visualized on the tracts increasing from red to yellow, and there appear to be increases in diffusivity in the vicinity of the white matter lesion on the left. Pelletier found an overabundance of white matter lesions within the anatomically defined thalamocortical tracts. (Image courtesy of Daniel Pelletier)

Brain atrophy in multiple sclerosis

Elizabeth Fisher from the Cleveland Clinic discussed how current imaging techniques are being used to determine the relationship between brain atrophy, pathology, and disease progression. Brain atrophy is a useful measurement because it can be assessed using conventional MRI machines found in most clinics today. Furthermore, it is present early on in disease progression and correlates moderately with disease severity. The prevailing thought is that brain atrophy is what happens "after and around the lesion." Fisher highlighted several assumptions that should hold if current thinking is true: 1) MS brains with lower lesion loads should show less atrophy, 2) MS patients should lose just as much white matter as grey matter, and 3) within an MS brain, regions with higher lesion load should be the most atrophied.

Fisher highlighted the apparent disconnect between the lesion dynamics and whole brain atrophy: while lesions appear to come and go across imaging sessions there appears to be a unidirectional expansion of the ventricles, particularly towards the end of the disease. In addition, it has been observed in clinical trials of anti-inflammatory drugs that even if the lesion load is greatly reduced, atrophy is reduced only 20%–50% above placebo. This strongly suggests that there is tissue loss that occurs independent of the lesion formation, and it supports a suspicion of researchers like Fisher that there is some source of positive feedback in neurodegeneration related to tissue damage. That is, it appears that "tissue loss begets more tissue loss."

Her group and others are trying to understand the causal connection between lesions and brain atrophy by looking for correlations of brain atrophy with other MRI markers. She reported that T2 lesions moderately correlated with brain atrophy. Fisher emphasized the importance of distinguishing true atrophy that originates from pathological processes such as demyelination, axonal loss, and neurodegeneration from normal age-related volume loss and artifacts stemming from dehydration at time of imaging.

Metabolism in the MS brain

Earlier talks mentioned using metabolites as markers that reflect changes in brain tissue but Matilde Inglese of Mount Sinai Medical Center, is interested in asking what spectroscopy can tell us more generally about the metabolic status of the MS brain. For instance, lactate, the end product of anaerobic glycolysis, and glutathione, a strong oxidizer, are elevated and decreased respectively in the brains of patients with MS. These changes suggest that there is an abnormal oxidative status in the MS brain. Changes in oxidative status can result from increased energy demands or decreased oxygen availability. Evidence suggests a combination of both factors is at play in MS—studies have found reduced blood perfusion in the brain as well as the compensatory redistribution of sodium channels along demyelinated axons that drives up energy demand.

Metabolites can signify changes to brain structures and potentially provide mechanistic insight into what drives brain alterations in the MS brain. As mentioned in other talks, a reduction in N-acetyl-aspartate is believed to indicate poor axonal health. Inglese stated that NAA levels provide additional information concerning the energy status of neurons because its synthesis is dependent on mitochondrial metabolism. Decreased function of the mitochondrial respiratory chain—consistent with a shift from aerobic respiration to glycolysis—would result in lower NAA levels. NAA is reduced in acute lesions, chronic lesions, normal appearing white matter, cortical grey matter, and subcortical grey matter in brains of MS patients. Furthermore, decreases in NAA levels were shown to correlate with clinical disability scores better than did lesion load. Inglese stated that while NAA is commonly used as a marker of axonal integrity, the correlation of NAA levels and grey matter atrophy has been weak. This suggests that there is more to the story of NAA than axonal health, and its link to metabolic status may provide mechanistic insight into MS pathology.

As mentioned previously, another change that could shift metabolism to anaerobic is a reduction in oxygen availability arising from poor perfusion. In arterial label spin imaging, arterial water is labeled in a proximal inversion plane and then after a delay (allowing for blood to flow there) the image is taken at a distal area. Findings from these experiments show white matter hypo-perfusion not only in active MS lesions but also in the normal appearing white matter of MS patients. A study found that in MS patients there was a reduction in the consumption of blood oxygen as detected by T2-relaxation-under-spin-tagging MRI. Finally, current studies are attempting to look at intracellular sodium atoms—the second most NMR sensitive nucleus, yet one of the least abundant nuclei in the brain. Measures of intracellular sodium concentrations could provide insight into sodium ion homeostasis in MS patients. There is evidence that the concentration of sodium channels increases in MS lesions and experimentally demyelinated axons, and in the EAE mouse model it was shown that partial sodium channel blockade prevented axonal degeneration. These finding hint that demyelinated axons compensate for a loss in axonal conduction by increasing the concentration of sodium. In turn, this sodium increase leads to an increase of intracellular calcium that is toxic to the cell and results in axonal degeneration. Preliminary studies suggest that intracellular sodium is increased in brains of MS patients.

Imaging in the Clinic: Practical and Experimental Applications (Part II)

Speakers:
Bruno Stankoff, Pierre and Marie Curie University
Jerry Wolinsky, University of Texas Health Science Center at Houston

Highlights

• The PIB radioligand is a promising marker for myelin and may be a sensitive tool to monitor remyelination in patients with MS.
• Looking to the future, drug trials need to be thoughtfully designed to assess neuroprotective effects of drugs in MS for which acute and chronic, systemic mediators of tissue damage are so interrelated.

Look at the bright side: imaging neuroprotection and repair in the MS brain

Advancements in imaging techniques that sensitively and specifically quantify myelin are important not only for tracking demyelination in disease but also for looking for signs of improvement—the reappearance of myelin. Bruno Stankoff of Pierre and Marie Curie University emphasized that a tool that could effectively evaluate remyelination and repair must be specific, high resolution, reproducible, quantitative, sensitive, and available in a clinical setting.

Stankoff highlighted the ways in which existing methods for imaging myelin fail to meet these criteria. A decrease in magnetization transfer ratio is believed to reflect a loss of myelin content due to a depletion of the pool of protons that are associated with the rigid macromolecular structure of myelin. However, evidence that MTR is influenced by water content and neuroinflammation limits the specificity of this measure. Stankoff noted that researchers have proposed the use of diffusion tensor imaging (DTI) to measure myelin content- a loss of myelin being reflected as an increase in radial diffusivity because demyelination leads to a reduction of water movement along the length of axonal tracts. However, it has been reported that axonal injury alters radial diffusivity, confounding specific measures of myelin integrity. Finally, he highlighted mean water fraction (MWF) measurements that previously raised questions of sensitivity and availability, but new developments like Deoni's mcDESPOT show promise.

Stankoff then described his own work imaging myelin by positron emission tomography (PET) using fluorescent thioflavin derivatives that bind protein structures with multiple beta sheets, such as amyloid and myelin basic protein. Stankoff reported consistent standardized uptake measures for radiolabeled Pittsburgh compound (PIB) in a study of healthy controls. Signal was enriched in white matter compared to grey matter and could not be detected when probed in mice lacking myelin basic protein, arguing for specificity. Preliminary data from a pilot study in MS patients shows an approximately 22% decrease in [¹¹C]-PIB in white matter lesions compared to normal appearing white matter. At this point, the small sample size shows high heterogeneity across subjects, with some exhibiting as high as a 40% reduction and others showing virtually no change. Interestingly, across imaging sessions (mean duration between 10.6 weeks) Stankoff has observed examples of potential remyelination, loci showing an increase in PIB uptake >10%. PET imaging of myelin holds promise but quantification methods must be optimized.

Imaging to monitor therapies aimed at neuroprotection

It can be a humbling experience to step outside of one's research field to see how outsiders view it. This is what Jerry Wolinsky of the University of Texas Health Science Center at Houston did when he decided to look up the Wikipedia article for neuroprotection. Neuroprotection was defined as the effect of any molecule or therapy that protects the brain from neurodegeneration or brain injury. The article went on to list a large number of diseases for which there's great interest in developing neuroprotective treatments, none of which were MS. Wolinsky stated that for the time being, MS should be viewed as a special case for neuroprotection—the strict definition of neuroprotection refers to mechanisms that prevent the apoptosis, or programmed cell death, of neurons. It appears in MS that CNS damage is for the most part systemically initiated by immune responses and inflammation. Therefore, a treatment that minimizes damage to myelin and axons by reducing inflammation may be considered "protective" but not in the strict sense. Wolinsky believes that treatments focused exclusively on either the inflammatory or neurodegenerative aspects would fail in neuroprotection because the acute and chronic mediators of tissue damage are so interrelated.

What imaging measurements should be used to assess neuroprotective treatments for MS? Data from Sormani's research suggest that MRI can be used as a surrogate for relapse because drug treatments that effectively reduced annualized relapse rates and disability also showed parallel reductions in lesion activity on MRI scans. Wolinsky's lab has been looking at subtraction imaging of T2-weighted images in order to detect changes in lesion activity across scans (careful image registration is key!). Results showed that increased lesion activity predicted which patients would show the most brain atrophy four years later. Wolinsky pointed out that the frequency at which scans are collected is likely to be very important; it's easy to imagine that lesion activity counts could be missed, possibly explaining their finding that lesions that appeared to shrink or disappear in the subtraction image also predicted brain atrophy.

Subtraction imaging of T2-weighted images can detect changes in lesion activity over time. (Image courtesy of Jeremy Wolinsky)

Wolinsky showed data from a drug trial that reported what looked like the disappearance of T1 hypolesions, which at first glance seem to imply neuroprotection. However, Wolinsky stressed that MRI markers can't be looked at in isolation when trying to draw conclusions about tissue repair. He presented an example of a T1 black hole that seemed to resolve, but upon closer examination it was clear that nearby tissue also disappeared, indicating more profound atrophy and not regeneration at all. The recently concluded CombiRx drug therapy trial found that on average, all treatment groups of relapsing remitting MS patients had a 1% loss of tissue volume in the first year, two-thirds of which took place in the first 6 months. Other studies report that imaging measures of tissue destruction take 6–12 months from the time of insult to appear. Thus this early damage goes untreated because of the time elapsed between damage and the start of treatment after MRI findings. This suggests that the brain atrophy from the first 6 months of the CombiRx represents what was already set in motion before treatment onset that was potentially unpreventable. It will be important to recognize this confound when assessing whether or not a drug is neuroprotective.

Wolinsky believes it will be worthwhile to study whether treatments can limit increasing disability in progressive MS. Because of the duration of these trials—and with the best interest of the patients in mind—active comparator groups should be allowed. Wolinsky argues that functional definitions, such as Expanded Disability Status Scale (EDSS) scores should be used instead of clinical classifications that are not always stable (e.g., sometimes patients classified as primary progressive suddenly remit). Patients should be older (>34) so that they will segregate in their disease progression, with patients crossing different EDSS thresholds during the course of the trial, lasting at least 5 years. Wolinsky closed by suggesting that adaptive seamless trial designs should be used. In the first 6–12 months outcomes could defined by MRI-defined lesion activity. The decision to keep trial arms can then be based on atrophy measures. Clinical outcomes would be defined by EDSS measures. Finally, he described the characteristics of a good candidate drug for such a trial:  to be neuroprotective it should cross the blood brain barrier and should be administered on top of an anti-inflammatory platform in order to address systemic effects.