MS Patient, Ph.D.: MRI Reading: Art or Science?
Neuroradiologists are like economists: Everyone has an opinion, and no two readers will agree on what they see in exactly the same data.
In April, I had a gradual onset of pain and muscle spasms in my left shoulder and arm, covering the C3/4 through C6 dermatomes. That arm has always been my “problem arm,” with a hyper bicep reflex and wrist weakness. After about a week, the pain and spasms had reached an unmanageable level, and having failed lorazepam and acetaminophen/hydrocodone, I had to do a multiweek course of prednisone. The steroid tamped down the pain, but the spasms are still there, now, 3 months after they began. My entire shoulder and arm seize up when I move, walk, stand, sit, or just do nothing. Nothing in particular seems to trigger it, and no specific movement does. I have full range of motion in my neck.
As part of the workup, I had a cervical MRI. The physician who ordered it called me on a Saturday morning—my wedding anniversary, as it happened—unknowingly interrupting the one effort my husband and I had made in a long time to eat at a nice restaurant, alone, together. His news was that the MRI failed to find any structural problems that would explain my symptoms. I have mild to moderate degenerative changes in my C-spine, but they don’t involve cord impingement and are more prominent on my right side, while my problems are with my left arm. These changes have been stable since 2008, I know, and an earlier CT myelogram had already shown that they weren’t causing any problems. What I did have, he said, were two spots of enhanced cord signal at C4 and C6 that were “concerning for demyelination.”
Even after 6 years of knowing I’ve got a problem in my brain and spinal cord, that news still rattled me. It would rattle anyone. Because our insurance network allows us to email doctors, I emailed this M.D. later that day, requesting a copy of the report. He declined to send it, essentially telling me I wouldn’t be able to understand or handle it. I informed him that regardless, it was my report and he needed to provide it, so he eventually did. At the bottom, the reader had commented that follow-up MRI of the C-spine with T1-weighted STIR sequences was recommended.
Over the past few years, thanks to having had several MRIs and because of my medical editing work in neurology, I’ve become a bit of an MRI junky. I remembered that this year, a report had come out of a double-blind study showing that T1-weighted STIR sequences were better for identifying cervical cord images than T2-weighted images. In that study of 29 patients with MS, the T2 sequence showed no cord lesions at all in seven of the patients, even though the STIR sequence revealed “significant cord involvement” in the same patients.
I’ve come to realize in my various dealings with MRI reports that neuroradiologists are like economists: Everyone has an opinion, and no two readers will agree on what they see in exactly the same data. The gorilla in the image can go unseen even as minor details are recorded assiduously but inconsistently. So I now expect that what one neuroradiologist sees, another will not. What surprised me in this experience was how they would differ, in-house, on the best way to get the image in the first place.
After I had undergone my follow-up MRI, I stood next to the technician while she went over the image set to make sure everything looked good to send along to the neuroradiologist. I didn’t hear her say “STIR,” so I asked her where the STIR sequence images were. She replied that those weren’t part of the orders. Surprised, I emailed the neurologist who had recently informed me that my gait problems weren’t PPMS, asking her why the STIR study that the original reader had recommended wasn’t part of the orders. She replied that a different neuroradiologist had sent the orders and that he said STIR was not appropriate for visualizing lesions in the cord. I know it probably makes me the World’s Most Annoying Patient Who Thinks She Knows Things, but I sent along a link to the most recent study and another retrospective study (using 3-T MRI) that suggest otherwise.
Meanwhile, the T2-weighted study they had done showed no lesions in the cord. I’m just a developmental biologist: I can read most papers in biology and medicine and understand them, but I don’t understand which study or reader interpretation is valid. Do I go with the results of the first MRI that wasn’t MS-specific but that identified two separate foci in the cord of “concern for demyelination,” areas that are precisely relevant to where I have issues and would also explain my Lhermitte’s? Or do I go with the second one, done with a sequence that evidently misses a pretty large percentage of spinal cord lesions completely and that found no lesions? Does it matter? Going or not going with either one doesn’t change much, practically speaking.
Because he requested it, I followed up with the M.D. who called me on my anniversary. He reviewed my latest MRI with me. The degenerative changes are slightly worse on my right than my left, and he badly wanted me to have problems on my right side. In fact, he was so insistent that it had to be my right side that it was almost comical. It’s not on my right. It’s my left.
He also wanted me to have problems only along one dermatome. I can’t help it if the spasm, tingling, and pain extend from where my neck meets my shoulder all the way into my thumb and index finger. Indeed, he seemed frustrated at being unable to neatly attribute my shoulder and arm problems to the C5 nerve root where there might be some narrowing. I have a positive Babinski response in my right foot now, but he didn’t comment on that.
I’ve been told repeatedly in my encounters with M.D.s in this insurance network that they can’t take my word for my medical history but have to have medical records. So I brought in a copy of MRI report summaries from my previous MRIs of my cervical spine, primarily because they have stayed the same for 6 years in terms of the degenerative changes. I thought that comparison would be useful. He took it, looked at it, and then said, “We can’t take copies of other readers’ reports. We don’t use outside readers,” and handed it back to me.
It’s a familiar refrain: No one trusts anyone else’s interpretations. I find it interesting that neuroradiology reading is so much an art and so little a science that no two neuroradiologists trust each other’s conclusions. Our insurance network’s own in-house readers evidently can’t even agree on the right way to image a cervical spine, much less on image interpretation. At this point, that leaves me stuck, yet again, in limbo. So in a sense, nothing has changed, except, I guess, I passed another life milestone during all of this: a wedding anniversary.
Key open questions
What is the best imaging sequence for cervical cord imaging?