Do patients with neurologically isolated ocular motor cranial nerve palsies require prompt neuroimaging?

Abstract

Neurologically isolated ocular motor (third, fourth, and sixth) nerve palsies present frequently to neuro-ophthalmologists. In older patients, microvascular ischemia is a common cause, and most patients recover spontaneously within several months. However, some patients have a more serious cause, including life- and vision-threatening conditions, such as expanding aneurysm, pituitary tumor and apoplexy, and brainstem stroke. The issue of whether to obtain neuroimaging in these patients (subjecting them to added time, expense, and risk of intravenous contrast) or following them over time without scanning (with the risk of missing a potentially treatable disease) is a dilemma faced by every physician, compounded by the increased attention paid to resource allocation and rising health care costs. Two experts weigh in on this topic in this Point Counter Point, using the available evidence to support their point of view. Pro: Neurologically Isolated, Presumed Vasculopathic Ocular Motor Cranial Nerve Palsies Require Prompt Neuroimaging: Nicholas J. Volpe, MD The necessity, yield, and effect on ultimate outcome of an acutely performed magnetic resonance imaging (MRI) study in a patient with an isolated ocular motor cranial nerve (CN) palsy is a matter of continued debate. This controversy is fueled by attempts to prospectively determining disease or lesion frequency and, frankly, one that no amount of data, can ever resolve. The reader should consider that there is little debate that the “yield” of such studies is around 5% (1). There is a 1 in 20 chance (much higher if you include third nerve palsies) that such a study could lead to a diagnosis of aneurysm, stroke, or tumor. I ask the reader to simply ask: “What would you really want done if you were the vasculopathic patient with a sixth nerve palsy?” Although we live in an era of increasingly constrained resources and we all must participate in the control of costs by reducing unnecessary care and testing, I would suggest that this clinical scenario is not one in which I would recommend our subspecialty try to control health care costs. The decision to image or not is at least partially based on a long standing and historical practice pattern that is rooted, in my opinion, in a different era of neuroimaging. Recall the days when an astute neuro-ophthalmologist could save the patient an invasive low-yield procedure, such as an arteriogram or pneumoencephalogram, by diagnosing isolated ocular motor CN palsy on a presumed vasculopathic basis. Fast forward and now imagine a time in which neuroimaging, at least in the United States, is readily available and is ordered and used as a screening tool for symptoms and signs with a much lower frequency of identified pathology. We are in an era in which new treatments for strokes and brain tumors have been developed, and there have been a plethora of case reports of isolated CN palsies from life threatening and highly treatable diseases. Finally, added to that debate, prospective studies (1–3), identifying a significant prevalence of important and identifiable lesions, leads me to recommend that most or all of these patients, particularly those with third and sixth nerve palsies, warrant neuroimaging studies. There are several considerations that support this decision and, at the outset, we must acknowledge that this is a decision that must be made on an individual basis by clinicians assessing patients with a specific set of symptoms. It is clear that the prevalence of identifiable and, more importantly, treatable pathology where there is evidence that treatment is cost effective and definitely matters (i.e., malignancy) is low. However, the argument that the decision to order an MRI here is not cost-effective or not reducing risk falls short. There is uniform agreement that both complete and partial third nerve palsies should be imaged because of the significant frequency of life-threatening aneurysms. For the fourth and the sixth nerve palsy, the majority of which are due to an ischemic or demyelinating process have an excellent prognosis for spontaneous recovery over 3 months. In these patients, MRI is unlikely to yield meaningful results that affect patient management. However, I do not agree that reduced testing in neuro-ophthalmic practice is our moment to be fiscally responsible. The words of Johnston and Hauser (4) urging restraint in contributing to rising health care costs are compelling and factual. Although not specifically studying this group of patients, another study found that neuro-ophthalmologists' decision to order imaging studies is cost effective in patients with CN palsies (all comers, not just older patients) (5). The advantages of identifying these disorders sooner rather than later are myriad and include improved clinical outcome, patients' psychosocial benefit, and avoiding the perception that appropriate diagnosis was delayed by a clinician's ignorance. In addition, other common neuro-ophthalmic scenarios in which MRIs are commonly ordered have a similarly low yield. For instance, unexplained facial or eye pain or unexplained, nonprogressive, unilateral optic atrophy are conditions shown to have a similar lower yield on neuroimaging studies even when ordered by an expert (6,7). Although I am aware of the societal fiscal issues related to responsibly caring for our population, that should never interfere with an individual physician's decision to order testing for an individual patient. There is a long list of tests that we order on our patients, also with low yield although often with high stakes, that have never been proven to be cost effective. Similarly, we have adopted many “expensive” behaviors (e.g., serial optical coherence tomography in various optic neuropathies) that have not been proven to impact patient outcomes, but still remain an important part of our tool box and assist the clinician trying to make the best possible decision for an individual patient. The studies performed by Chou et al (2), Murchison et al (3) (single center), and Tamhankar et al (1) (multicenter) have attempted to guide clinical decision making. Each study has its potential flaws and biases and, in fact, the reader could likely use any of the studies to support the decision to image or not to image. In each of these studies between 1% and 16% of patients (depending on which study and which subset of patients analyzed), all evaluated by neuro-ophthalmologists and presumed to have vasculopathic ocular motor cranial neuropathies were found to have various conditions ranging from stroke, cerebral hemorrhage, pituitary apoplexy, aneurysm, and benign and malignant brain tumors. These important conditions were detected in patients who would have otherwise been thought to have vasculopathic CN palsy. Furthering the case for imaging, creating restricted guidelines might lead to a false sense of security and reassurance in the hands of a non-neuro-ophthalmologist who is not as experienced in recognizing the subtleties of cases that demand further attention. We can assume that the experienced neuro-ophthalmologist can recognize patterns of human disease and variances that will inform his or her decision to image (the tempo at which symptoms develop or some subtle-associated symptom or examination finding). However, a blanket statement about isolated CN palsies in vasculopathic patients not needing neuroimaging studies, unless they fail to recover, is likely to provide nonexperts with a false sense of assurance in clinical situations that are really not straightforward. Given the relatively low access to neuro-ophthalmologists, it seems prudent to recommend imaging studies in the setting of all patients with isolated ocular motor CN palsy. It is not appropriate for us to conclude, through generalization, that treating some of these patients for their identified condition “would not have mattered” or has not been proven to improve outcomes. I think that the reader would certainly want to know whether they had a brainstem stroke causing a sixth nerve palsy vs a normal MRI and presumed vasculopathic palsy. Similarly, the presence of pituitary apoplexy or a small structural lesion such as an aneurysm is high-stakes diagnoses that require prompt attention. As modern imaging studies evolve, and our ability to image CNs in a more detailed fashion increases, the argument that we might gain a better understanding of the mechanism of these palsies, or discover previously unknown abnormalities, may also prove to be important. In some cases, a false sense of security, both for the doctor and for the patient, can come from a clinical course of spontaneous resolution. Spontaneous resolution by no means excludes important pathologies. Certainly, for most brainstem strokes, the clinical symptom of an isolated cranial nerve palsy will spontaneously resolve. Spontaneous resolution of ocular motor CN palsies has been described in the setting of structural skull-based lesions (8). These lesions may not require immediate or urgent treatment or might declare themselves at a later date, but ultimately in this era of available treatments and patients perception of “sooner would have been better,” most would argue for prompt diagnosis and subsequent decision making in the setting of skull-based neoplasms. Ultimately, we agree that each clinician must decide what age, what risk factors, what clinical scenario (pretest likelihood), and what access and availability issues affect his or her decision to image patients. My work and that of others on this topic is meant to guide this decision, and I agree that there is insufficient data to force or mandate a practice pattern. However, one must acknowledge, based on the reported studies, that there is a small but significant yield from these diagnostic tests and that clinically patients with structural lesions are not easily distinguished from their counterparts with the more innocent vasculopathic palsy. Clinicians must decide, based on the case in front of them, whether the information obtained from an MRI could matter to this particular patient. If one considers the various circumstances in neuro-ophthalmic practice where MRIs are ordered, isolated ocular motor CN palsies should be added to the list of where neuroimaging could prove helpful. I believe that the yield is high enough to allow for this to be “cost effective” for an individual patient. Con: Patients With Neurologically Isolated, Presumed Vasculopathic Ocular Motor Cranial Nerve Palsies Do Not Require Initial Neuroimaging: Andrew G. Lee, MD I do not disagree with my colleague that neuroimaging has an important role in the evaluation of many ocular motor cranial neuropathies; instead it is my contention that the value of such imaging in neurologically isolated, non-third nerve-related, and presumed vasculopathic ocular motor CN palsies is limited and ultimately not cost effective. My arguments against neuroimaging in this setting include: 1) high cost, 2) low yield in this clinical setting, and 3) the lack of significant impact on the treatment plan or prognosis for the majority of patients. High Cost Cranial MRI is expensive, and a complete head and orbit study with fat suppression and intravenous contrast adds a significant additional cost. In an editorial on the cost of stroke imaging in Annals of Neurology, Johnson and Hauser (4) wrote that “Neurologists can no longer ignore the headlines about rising healthcare costs. Setting aside the politics and the biases, it is undeniable that the costs of healthcare in the United States continue to rise at an alarming rate that will either bankrupt us or force us to ration. We have all seen the numbers. US healthcare costs accounted for 16% of total gross domestic product (GDP) in 2008, well above all other large countries, with France a distant second at 11% of GDP. Worse yet, healthcare costs are rising faster than GDP and, if unchecked (an impossibility), are predicted to bring down the US government and the competitiveness of our businesses.” Likewise, we as neuro-ophthalmologists also must face the reality of cost containment and the utilization and yield of neuroimaging. We as specialty leaders also need to stand firm in our conviction that our clinical judgment and pretest likelihood of disease (e.g., vasculopathic etiology) still counts for more than any test including neuroimaging. Murchison et al (3) prospectively studied 93 patients (older than 50 years) with acute isolated ocular motor cranial mononeuropathies. Although 4 of these 93 patients (4%) had lesions on MRI, only 1 (1%) had a lesion related directly to the cranial mononeuropathy. The total modeled cost of imaging spent for these 93 patients was $131,688 to determine an underlying cause in only 1 patient and in whom there was no ultimate change in treatment. These authors concluded that “it may not be medically necessary to perform MRI scanning on every patient with an isolated CN III, IV, or VI palsy.” In my opinion, it is very difficult to justify the high cost of negative neuroimaging for the majority of patients with an acute and neurologically isolated mononeuropathy secondary to ischemia to find the 1 treatable lesion. Low Yield Tamhankar et al (1) performed a prospective study to estimate the proportion of patients presenting with isolated third, fourth, or sixth nerve palsies of presumed microvascular origin vs other more concerning intracranial causes. These authors evaluated 109 patients, aged 50 years or older, with acute isolated CN palsies all of whom had an MRI of the brain. Among these 109 patients, there were 22 who had palsies of the third nerve, 25 of the fourth nerve, and 62 of the sixth nerve. A cause other than presumed microvascular ischemia was identified in 18 patients (16.5%; 95% confidence interval, 10.7–24.6). As expected, the presence of vasculopathic risk factors (e.g., diabetes, hypertension, hypercholesterolemia, coronary artery disease, myocardial infarction, stroke, and smoking) was significantly associated with a presumed microvascular cause (P = 0.003, Fisher exact test) but these vasculopathic risk factors were also present in 11/18 patients (61%) with other causes. In the group of patients who had vasculopathic risk factors only with no other significant medical condition, 8/80 (10%) were found to have other causes. These alternative etiologies included midbrain infarction, neoplasm, inflammation, pituitary apoplexy, and giant cell arteritis (GCA). If the patients with third nerve palsies and GCA were excluded, then the prevalence of other causes of fourth and sixth nerve palsies was 3/64 (4.6%). The authors concluded that for “acute isolated ocular motor nerve palsies, a substantial proportion of patients had other causes, including neoplasm, GCA, and brain stem infarction” and that “brain MRI and laboratory workup have a role in the initial evaluation of older patients with isolated acute ocular motor nerve palsies regardless of whether vascular risk factors are present.” I believe that a strong case can be made using the same data from this report to conclude the opposite: those patients with neurologically isolated ocular motor cranial neuropathy should not undergo neuroimaging because the yield is low overall (i.e., 16.5%) for finding an alternative etiology. If third nerve palsy and GCA are removed from the cohort, the yield drops to 4.6%. I would argue that obtaining a brain MRI in these patients actually might harm the patient because a CT angiogram (CTA) or magnetic resonance angiography (MRA) study are typically the first line studies for third nerve palsy and MRI of brain in GCA would be expected to be normal. Thus, ordering an MRI in these settings might delay diagnosis, provide a false sense of security for etiology, and add the cost neuroimaging. There is already considerable clinical evidence to support the concept that a third nerve palsy does not carry the same clinical significance for harboring a serious underlying intracranial problem as an isolated fourth or sixth nerve palsy, and that combining all 3 ocular motor cranial nerve palsies is not a valid method of imaging yield for all cranial mononeuropathies. In my clinical practice and in the literature, the most common etiologies for fourth nerve palsy are congenital, traumatic, and ischemic, and the yield for neuroimaging in this setting is extremely low. Third nerve palsy, in contrast, especially incomplete palsies (e.g., divisional palsies) or pupil-involved palsies, are especially dangerous, and the presence of vasculopathic risk factors would not dissuade me from imaging for aneurysm (e.g., CTA or MRA) even when neurologically isolated. Likewise, a sixth nerve palsy, even in isolation, is potentially more dangerous than a fourth nerve palsy and may be due to a clival or cavernous sinus lesion. In addition, sixth nerve palsy is also of concern because of the possibility of a nonlocalizing sign of increased intracranial pressure. Lack of Impact on Treatment Plan or Prognosis Another prospective study (9) evaluated the use of MRI in 43 consecutive patients with acquired sixth palsies, and causative lesions were identified in 27 of 43 patients (62.8%). However, the median age of these patients with lesions on initial imaging was 43 years compared with 56 years in the patients with normal MRI results, and individual case details were lacking to determine whether a “reasonable and prudent” neuro-ophthalmologist would have judged the sixth nerve palsy to be isolated or presumed vasculopathic. Another prospective study (2) evaluated ocular motor cranial mononeuropathies and included only patients older than 50 years. Nine of 66 patients (13.6%) had a cause other than a microvascular etiology but unfortunately included 9 patients with pupil-involving third nerve palsies and 10 patients with partial third nerve palsies (who I would have imagined [e.g., CTA and MRI/MRA]). In fact, 2 of these patients were found to have aneurysms. Thus, the actual diagnostic yield for neuroimaging was 7 of 66 patients (10.6%). The most frequent finding in this study was meningioma (3/7 patients), and it is unlikely that these cranial neuropathies would have resolved spontaneously. A delay in diagnosis during an observation period likely would have had no impact on the final treatment or prognosis for meningioma. Excluding these patients, the diagnostic yield would be mere 4/66 patients (6.1%). Based on 1) the high cost of neuroimaging, 2) the low yield of imaging in the appropriate clinical setting, and 3) the lack of change based on such imaging on the treatment plan or the prognosis for the majority of patients, I do not believe that neuroimaging is warranted for every vasculopathic patient with an acute, neurologically isolated, ocular motor cranial neuropathy. Ultimately, patients and doctors should be free to make an informed choice based on pretest likelihood of disease and an individual risk-to-benefit decision analysis and I do not believe that neuro-ophthalmologists should be forced into any single evaluation paradigm that mandates expensive and potentially unnecessary testing. Such unnecessary testing, mandated by a published paradigm, risks bankrupting our country's health care finances. Volpe Rebuttal A palsy is not always just a palsy, and I believe part of the problem is that we are analyzing third, fourth, and sixth nerve palsies together. I share my colleague's concern about the high cost, low yield, and a lack of change in management as highly intuitive and largely evidence based. This argument could guide decisions about large populations and the allocation of limited health care resources as a model of health care delivery. Isolated ocular motor CN palsy is not a clinical setting in which an MRI will uniformly lead to a high-yield diagnosis or one that would usually impact a patient's management with a specific result. That being said, I would urge the reader to think again about what they would want done if it were themselves and whether in their minds there is a value to a 5% identification rate of conditions such as brainstem stroke, pituitary apoplexy, or a skull-based neoplasm or vascular lesion. In addition, I wonder if we are only disagreeing about fourth nerve palsies. My colleague specifically writes about his own clinical practice and informs the reader that third nerve palsies are almost always imaged with MRI and MRA or CTA. He goes on to state that sixth nerve palsies in isolation are potentially more dangerous because of clival or cavernous sinus lesions and the possibility of sixth nerve palsy being nonlocalizing due to elevated intracranial pressure. Are we willing to concede that third and sixth nerve palsies warrant imaging? I would add to the sixth nerve palsy argument that the yield has certainly shown to be high enough in people younger than 50 years (seemingly an arbitrary cutoff). Perhaps 55, 60, or 70 years is the new 50 for this diagnosis? In the end, I do not recommend doing yet another multicenter prospective trial to determine the yield of imaging studies on elderly patients with isolated fourth nerve palsies. For these patients, I am willing to concede! Lee Rebuttal Dr Volpe asks the reader to consider “What would you really want done if you were the vasculopathic patient with a sixth nerve palsy?” I would argue that the answer is “it depends.” Neuroimaging is not always benign, and these studies can uncover incidental and unrelated pathology (e.g., frontal or falx meningioma) that can increase rather than decrease patient anxiety and may lead to further unwarranted studies. In addition, gadolinium contrast in MRI has some risks especially in patients with renal failure (e.g., nephrogenic systemic dermatopathy) that might preclude appropriate use of contrast or produce unwanted side effects. Finally, if the question really is “What would the readers of our journal really want to do if they developed a vasculopathic sixth nerve palsy?” then I would imagine that they would answer: “I would like to make my own decision after appropriate informed consent and an individualized analysis of risk, benefit and cost.” The clinician's role is to provide sufficient information to the patient to make a decision and not to make the decision per se unless specifically asked to do so. CONCLUSIONS Every patient presenting with an isolated ocular motor CN palsy potentially harbors a serious underlying cause, and the diagnosis of a treatable condition can be vision or life-saving. Before ordering neuroimaging, we all perform a formal or informal risk stratification assessment, based on a variety of factors, such as age, vasculopathic risk factors, associated clinical findings, the pretest probability of identifying a causative lesion, and the time and cost of the scan. It is also worth acknowledging that although our neuroimaging capacities are constantly improving, the ever-increasing sensitivity is not necessarily accompanied by increasing specificity, so an MRI might not show a causative lesion, but uncover unrelated findings, or “incidentalomas” that require further testing. Although we tend to lump third, fourth, and sixth nerve palsies together, our experts argue convincingly that not all CNs are created equally, and we can use our knowledge about anatomy and pathophysiology to identify the patients at the highest risk for serious underlying pathology, with the highest yield for neuroimaging.