Clinicopathological Conference: A Deadly Cause of Seizures in a 67-Year-Old Alcoholic
Casey Z. MacVane MD, MPH,
From the Department of Emergency Medicine, Maine Medical Center (CZM), Portland, ME; and the Department of Emergency Medicine, SUNY Downstate Medical Center & Kings County Hospital (CID), Brooklyn, NY.
From the Department of Emergency Medicine, Maine Medical Center (CZM), Portland, ME; and the Department of Emergency Medicine, SUNY Downstate Medical Center & Kings County Hospital (CID), Brooklyn, NY.
A 67-year-old male presented via emergency medical services (EMS) to the emergency department (ED) of an urban academic medical center with a chief complaint of seizure. The patient’s wife called EMS after finding him on the floor that morning with generalized, shaking movements. He was incontinent of urine and stool. She last saw the patient the previous evening and reports that he had been in his usual state of health. He drank several bottles of wine the day before, which was not unusual for the patient. She stated that he had never had a previous seizure.
EMS reported that when they arrived on scene, the patient appeared postictal. His blood glucose level was 200 mg/dL. En route to the ED the patient had a subsequent tonic–clonic seizure, which lasted approximately 1 minute and was aborted with intravenous midazolam.
The patient’s past medical history, as provided by his wife due to the patient’s obtundation, included non–insulin-dependent diabetes, hypertension, hyperlipidemia, and gastroesophageal reflux disease. His regular medications included metformin, furosemide, atorvastatin, pantoprazole, and duloxetine. The patient was retired and living with his wife. He was a former pipe smoker and had a long history of alcohol abuse. At the time of his presentation, he had continued to drink heavily but had never experienced withdrawal seizures. He had no recent trauma and had not been complaining of headache, fever, chills, nausea, vomiting, chest pain, or shortness of breath.
On presentation, the patient’s vital signs were: temperature 38.5°C, heart rate 112 beats/min, respiratory rate 20 breaths/min, blood pressure 132/79 mm Hg, and oxygen saturation 100% on 15 L of oxygen via nonrebreathing mask. On physical examination he was noted to be morbidly obese and had sonorous respirations. His head was atraumatic. Examination of the patient’s eyes revealed symmetric, reactive pupils with both eyes deviated to the right. His neck was supple and without lymphadenopathy. His lungs were clear and without wheezes, crackles, or rhonchi. His cardiac examination was regular and tachycardic with normal heart sounds and no appreciable murmurs. His abdominal examination was unremarkable. Brown, heme-negative stool was noted on rectal examination. Extremities had 1 + pitting edema bilaterally. His skin had bilateral lesions on the plantar aspects of his first metatarsophalangeal joints. The lesion on the right had purulent drainage. Somnolence was noted upon neurologic examination. He was moving all extremities spontaneously but not following commands. He occasionally groaned in response to painful stimuli without localization.
Shortly after arriving in the ED, the patient had another generalized, tonic–clonic seizure. He was given 2 mg of lorazepam intravenously without cessation of the seizure. He was intubated for airway protection, and a loading dose (20 mg/kg) of fosphenytoin was administered intravenously. A postintubation chest x-ray was obtained (Figure 1). An electrocardiogram (ECG) was also obtained (Figure 2).
The patient’s complete blood count established the presence of a leukocytosis with 17.2 × 109/L, hemoglobin 9.3 g/dL, hematocrit 27.4%, platelets 218 × 109/L, with a differential of 85% neutrophils and 12% bands. A chemistry profile revealed sodium 125 mEq/L, potassium 4.8 mEq/L, chloride 85 mEq/L, bicarbonate 25 mEq/L, blood urea nitrogen (BUN) 27 mg/dL, creatinine 1.2 mg/dL, and glucose 228 mg/dL. Hepatic panel was notable for aspartate transaminase 78 U/L, alanine transaminase 46 U/L, and alkaline phosphatase 116 U/L. Cardiac enzymes included a creatine kinase, total of 128 U/L, creatine kinase, MB isoenzymes 9.3 ng/mL, and troponin T 0.18 ng/mL. Coagulation parameters revealed an international normalized ratio of 1.4 and a partial thromboplastin time of 38 seconds. Urine and blood cultures were collected and urinalysis was unremarkable.
A head computed tomography (CT) was obtained (Figure 3). Subsequently, a lumbar puncture was performed and clear fluid was obtained. Cerebrospinal fluid (CSF) studies demonstrated a white blood cell (WBC) count of 280 × 106/L with 84% neutrophils and 0% lymphocytes, red blood cell count of 23 × 106/L, protein of 83 mg/dL, and glucose of 132 mg/dL.
The patient’s repeat blood pressure was 90/51 mm Hg. A rapidly infused bolus of 2 L of normal saline was administered, and the patient was started on piperacillin/tazobactam and vancomycin. The intensive care unit (ICU) physician was contacted for admission. At this point a diagnostic test was performed, leading to the diagnosis.
“Alcoholics and frequent flyers were put on the earth to make emergency physicians look silly.” Corey Slovis, MD, chair of emergency medicine at Vanderbilt University Medical Center, uttered these words in my not-so-distant-past at a national conference. I chuckled knowingly when he said it. I was not laughing alone.
Dr. Slovis’s were the first words that came to me when I initially read this case. I was facing unraveling the differential and hopefully finding the correct diagnosis in this elderly male “career drinker” with a first-time seizure who has now become hypotensive.
At first pass, the case seemed easy. The patient was febrile, with a moderately high leukocytosis, a bandemia, a left shift noted on WBC differential, altered mental status, new-onset seizures, and a positive lumbar puncture. This was a clear case of meningitis or encephalitis, and it would be difficult to tell these two diagnoses apart clinically at the time. I had solved this diagnostic dilemma in just under 6 seconds. However, CPC cases are never so easy. I was going to need to delve deeper. Like all great detective stories, it is often difficult to determine what is a clue and what is a red herring. Every piece of evidence must be examined, weighed, measured, and placed into context. History is everything.
All the history in this case was provided by a surrogate historian. This can often complicate the case in several ways; the surrogate does not always know the complete history because the patient has hidden some part of his or her social or medical history from the surrogate. The other significant issue with surrogate histories is that the surrogate can consciously or subconsciously steer the history and provide significant bias. Also, surrogates will also sometimes withhold information that is embarrassing to the patient or the surrogate.
However, in medicine, we only have the history obtained, so we must take it for what it is worth, realizing it can be “filtered,” incomplete, or frankly wrong. When considering this patient’s history, there are several key points that seem to jump out at us. The patient was in a normal state of health until the day of presentation and was not ill in the preceding days. This makes the illness acute and rules out a slow, chronic disease. The patient also had wine yesterday that possibly could have been contaminated or have toxins in it. The patient seemed to have a generalized clonic seizure with a classic postictal phase. There was no previous history of any seizures in this patient, although other previous ethanol withdrawal syndromes were not mentioned.
The seizure repeated several times and initially responded to benzodiazepines. With a blood sugar of 200 mg/dL, hypoglycemia was ruled out as a cause of his seizures and altered mental status. The past medical history does not reveal a common cause of new-onset seizures except for the possibility of an intracranial hemorrhage caused by hypertension. In addition, long-standing diabetes mellitus can be a cause of immunosuppression and infection due to microvascular disease.1 This microvascular disease can induce peripheral ischemia and the decreased function of white blood cells during hyperglycemic states.
His medications are a little more helpful. The patient is on duloxetine, an atypical antidepressant that blocks the reuptake of biogenic amines in the central nervous system. Any medicine that increases the availability of these biogenic amines can theoretically cause seizures by increasing the concentration of excitatory neurotransmitters in the brain. In fact, duloxetine has been implicated in seizures. It has also been associated with hyponatremia and, specifically, the syndrome of inappropriate antidiuretic hormone secretion (SIADH).2 Our patient was noted to have hyponatremia.
The physical examination was significant for several findings. The patient was febrile and tachycardic with a normal blood pressure and blood oxygen saturation. He was postictal with sonorous respirations, but was morbidly obese. The patient had a gaze preference to the right side, which may have been related to the seizures or intracranial pathology. He had a normal cardiac and pulmonary examination without adventitious breath sounds, cardiac ectopy, or murmurs. The extremity examination showed peripheral edema with some vague plantar lesions and an ulceration with probable infection on the right foot. The neurologic examination was vague. The Glasgow Coma Scale (GCS) was unable to be calculated, but the patient was moving all extremities and had altered mental status. Subsequently, the patient began to actively seize again. The seizures terminated with fosphenytoin after a repeat dose of benzodiazepines failed to end the seizure. This suggests that there might be a discrete nidus of epileptogenic activity in his brain such as an intracranial hemorrhage, tumor, or abscess.
The patient subsequently dropped his blood pressure and was given crystalloid infusion. With the obvious infective picture and hypotension, the emergency physician assumes the patient to be in septic shock and treats with broad-spectrum antibiotics including coverage for Streptococcus pneumococcus and methicillin-resistant Staphylococcus aureus (MRSA).
The laboratory data are full of abnormalities. The WBC count was grossly elevated with a significant left shift and bandemia. These values, along with the fever, suggest systemic infection. The patient also had hyponatremia and hypochloremia of unknown cause. The serum chemistry was helpful and showed a BUN:creatinine ratio of 22.5, which suggested dehydration. The calculated anion gap was 18, which is elevated. Of note, the patient’s albumin was 2.6, which is of unknown significance, but will affect the anion gap calculation. The patient’s cardiac markers showed a troponin level that is 600% of normal, suggesting an acute myocardial infarction. The patient was also noted to have mildly elevated liver enzymes, but this is expected in a chronic ethanol abuser.
The ECG was consistent with myocardial ischemia and showed ST-segment depression in the lateral leads. The baseline ECG for this patient was not provided so specific morphology analysis is limited.
The lumbar puncture results were consistent with CNS infection but not conclusive for any specific infective agent. Usually bacterial meningitis of the severity to cause seizures and septic shock will present with a WBC count of over 1000 cells/hpf.3 The cell counts on this patient were consistent with bacterial, fungal, viral, or mycobacterial sources, but the differential of the white cells suggests that viral and mycobacterial causes were less likely. The chest radiograph showed no focal consolidations or obvious causes for hypotension.
The head CT showed no intracranial hemorrhage, mass, or edema, which narrowed the differential of altered mental status and seizures significantly. The CT did show diffuse atrophy, possible left frontal sinus disease, and a small area of high attenuation in the right front parietal cortex. This lesion was a possible nidus for the seizures but had no surrounding edema to suggest infection or tumor.
The Principle of Parsimony, also known as Occam’s razor, dictates that we try to find a single unifying diagnosis to account for all of the patient’s signs and symptoms.4 To tie the issues together, it is useful to be very clear with the problem list. The four main issues in this patient are sepsis, a wide anion gap, new-onset seizures, and an atypical pleocytosis on CSF examination. All of these findings suggest meningitis or encephalitis, but the level of pleocytosis and the lack of meningismus on examination are atypical for bacterial meningitis. While the atypical pleocytosis and lack of meningismus do not rule out bacterial causes, the astute clinician will not close the differential diagnosis prematurely and will search for another possible cause.
Let us look at these four issues separately. Sepsis is defined as two systemic inflammatory response syndrome markers and a proven or suspected source of infection.5 Our patient potentially meets criteria of septic shock depending on his response to fluid resuscitation. The obvious source of sepsis is meningitis, but a systemic bacteremia with meningeal seeding must not be excluded. Based on the description of the right foot ulcer, we cannot ascribe a systemic response to the soft tissue infection there. However, this may be a source of a virulent organism such as MRSA seeding the bloodstream.
The widened anion gap has a large differential in itself. In the serum, the number of cations and the number of anions must be the same. The main measured serum anions are bicarbonate and chloride, while the main cation is sodium. The anion gap is usually approximated by subtracting the bicarbonate and chloride from the serum sodium. The normal solution to this equation is 8 to 12. This gap accounts for various organic acids in the blood and the serum albumin; these are the “unmeasured anions.”6 Figge et al.7 pointed out that albumin accounts for almost the entire anion gap, and thus hypoalbuminemia can cause a decreased gap. By Figge’s correction method, our patient’s gap was actually 28, much higher than the initially calculated figure of 18, and significantly above normal.
The common causes of an elevated anion gap are few; they are easily recalled with the mnemonic KULTS: ketones (from any source), uremia, lactic acid (from any source), toxic alcohols, and salicylates. The most likely causes of this patient’s increased gap are toxic alcohols and lactic acid. Lactic acidosis is very common in sepsis and septic shock, which we assume is present in our patient. Lactic acid is also produced by seizure activity. Alternatively, the ingestion of toxic alcohols by chronic alcoholics has been well documented as a cause of widened anion gap.8 Not realizing its deleterious effects, chronic alcoholics often ingest a toxic alcohol when they do not have another source of ethanol. Furthermore, toxic alcohols, particularly methanol, are a common contaminant in homemade alcohols. The degradation products from the body’s metabolism of toxic alcohols add to the amount of organic acids acting as anions in the serum. These acids are reflected as a widened anion gap on serum chemistry. One toxic alcohol, ethylene glycol, causes seizures in addition to a widened anion gap.9 Diethylene glycol, which has been reported as a contaminant in homemade wines, is also known to cause seizures.10 There are too many expected sources of lactic acid in this patient to discount the possibility that lactate may be the cause of the widened anion gap.
The presence of new-onset seizures actually helps us quite a bit with this patient. The common causes of new-onset seizure are relatively few. There is no clear history or physical examination findings consistent with trauma, so we can discount a head injury as a cause for this patient’s symptoms. Further, there is a relatively normal head CT given for this patient, so we can assume trauma is an unlikely cause for the patient’s presentation. Also, the head CT shows no evidence of intracranial hypertension, hemorrhage, or hematoma. This also takes a common cause of new-onset seizures out of the differential diagnosis. There is no evidence of metabolic causes for seizures either. Hypoxia and hypoglycemia are common causes of new-onset seizures, but we do not see evidence of those issues in this patient. Neurovascular disease and stroke can cause seizures, but the head CT shows no evidence of severe neurovascular disease or old stroke as an epileptogenic nidus. This leaves two broad categories for the cause of the seizures in this patient and we can be fairly certain that the cause is from one of these two groups.
Seizures are a very common presentation of toxicologic exposure. Hundreds of commonly used medications can cause seizures in therapeutic or supratherapeutic doses.9 There is little evidence here of exposure to toxicologic agents other than the possible contaminated ethanol. If the patient had been exposed to another toxin, then it has such subtle historical clues and physical findings as to make it impossible to diagnose.
The patient’s medications include metformin and duloxetine. While antihyperglycemics are associated with seizures via a hypoglycemia mechanism, metformin should not cause true hypoglycemia. Sulfonylureas will cause increased insulin secretion and therefore hypoglycemia and possibly seizures. As the patient was not reported to be on any sulfonylureas, nor found to be hypoglycemic on the serum chemistry, we can discount this medicine as a cause of the seizures. Duloxetine is an uncommon cause of seizures. Duloxetine decreases the seizure threshold by increasing the availability of biogenic amines like norepinephrine in the brain; therefore, duloxetine is a possible cause of seizures in this patient. As was mentioned before, ingestion of other alcohols besides ethanol can cause seizures. Withdrawal seizures, or “rum fits,” could explain the seizures in this patient, and the timing of withdrawal seizures would fit this patient’s presentation. However, although toxicologic exposures could cause the patient’s seizure, it would not explain the three other major clinical abnormalities and septic findings. Thus, we must discount toxicologic exposures or ethanol withdrawal as causes for the seizure at this point with the exception of the possible causes of duloxetine and ethylene glycol.
Another common cause of seizures that is more likely in this patient is infection. Many infections can cause the clinical picture of sepsis that we see in this patient, but only a few cause seizures. Severe pneumonias can cause seizures through a hypoxic mechanism, but we do not see evidence of profound hypoxia in this patient, nor is there an abnormal chest x-ray. Shigella has been known to cause bacteremia and seizures. We have no history suggesting severe diarrhea, so this would be unlikely. As mentioned before, meningitis and encephalitis would be very common causes of sepsis and seizures, so these diagnoses will remain high in the differential. The other infectious diagnosis that we must consider is seeding from another source. It is also common for bacteremic infections to seed the meninges and provide an epileptogenic focus for the seizures.11
The pleocytosis seen on CSF examination directly suggests an infection of the meninges or brain parenchyma. The predominance of neutrophils and a total WBC count of 280 suggests early bacterial meningitis or encephalitis. Meningitis secondary to viral or mycoplasma organisms would normally cause a lymphocyte predominance, which was not seen in this patient. Fungal meningitis could cause this type of pleocytosis, but is usually associated with significantly immunocompromised states, which was not given as history in this patient. The elevated serum glucose clouds the interpretation of the CSF glucose. By comparing the serum and CSF values, we see that the CSF glucose is 60% of the serum glucose, which is consistent with several causes of meningitis. Therefore, the CSF glucose level is not helpful in narrowing our differential of the patient’s pleocytosis. We are left with these CSF abnormalities being caused by early bacterial meningitis and encephalitis, and it is difficult to differentiate between these infections based on the CSF results. Bacterial meningitis and encephalitis are devastating infections and account for all four of the major features we identified. Meningitis secondary to Neisseria meningiditus is often associated with cutaneous lesions, which our patient had, but these are usually described as purpuric and not ulcers. There are no classic findings on head CT of early bacterial infections, so the head CT is not helpful for assessing bacterial meningitis. However, it is possible that the cutaneous lesion is a source of bacteremia leading to sepsis and meningitis.
This brings us to the last possible diagnosis. Bacterial endocarditis explains this patient’s entire presentation. Whenever there are multiple organ systems involved in an infective process, bacterial endocarditis must be considered. Perhaps the lesion on the foot is much more important to this case than we originally thought. It might be a source of bacteremia and endocarditis. Septic emboli from a cardiac vegetation could be showering this patient’s vascular system. The lesion on the head CT that we previously discounted might actually be a septic embolus that has seeded the meninges and is the nidus of the seizure activity. The lesion on the foot could be either a septic embolus from the heart, also known as a Janeway lesion, or a port of entry of the organism that has infected the cardiac valve. Janeway lesions are hemorrhagic microinfarctions seen in endocarditis and usually appear on the palms of the hands and soles of the feet. If cultured, they can show the causative organism.
Bacterial endocarditis is also noted to often cause the SIADH syndrome. This would explain the hyponatremia and hypochloremia seen on the serum chemistry. One of the serious sequelae of endocarditis is embolization to the vascular tree. A small embolus from the endocarditis vegetation could cause a microinfarct anywhere in the vascular system. It could cause the small cardiac infarction and resulting troponin elevation and ECG changes in this patient. An embolus to the kidney could seed the urine and cause infection seen on urinalysis. Additionally, an embolus to the brain in this patient could have caused the meninges to be seeded and given the pleocytosis seen on CSF examination.
Bacterial endocarditis is the diagnosis that best fits this patient’s main issues of sepsis, new-onset seizures, and an atypical pleocytosis on CSF examination. The tests that should be performed next are several sets of blood cultures and echocardiography. In any event, the patient should be treated with broad-spectrum antibiotics, benzodiazepines, and supportive monitoring.
This patient, who presented with a new-onset seizure, had findings of fever, bilateral foot ulcers, evidence of ischemia by ECG and cardiac enzymes, and an abnormal head CT. The diagnostic test that was performed was a transthoracic echocardiogram (TTE; Figure 4). This revealed an aortic valve vegetation, leading to a final diagnosis of infective endocarditis.
The patient was admitted to the ICU and remained hypotensive, requiring aggressive pressor support. His blood cultures grew out MRSA, and he remained on antibiotics. A subsequent MRI of his feet confirmed the presence of bilateral osteomyelitis necessitating amputations of his great toes. He developed multisystem organ failure, and thus was deemed not to be a candidate for aortic valve replacement. The patient expired on hospital day 18 after his family decided to withdraw continued support.
This patient presented with new-onset seizure, which in and of itself has a broad differential. The patient had a history of alcohol abuse, but early in the patient’s course it became clear that an alcohol withdrawal seizure did not explain this complex presentation. Evidence of systemic infection, including fever, tachycardia, hypotension, an elevated WBC count, and infectious CSF, were also present. A potential source for this patient’s infection was the bilateral foot ulcers found on physical examination. This patient also had evidence suggesting myocardial infarction with elevated cardiac biomarkers, including a positive troponin and anterior ST elevations on his ECG, with reciprocal, lateral ST depressions. His head CT revealed a focal lesion that was likely a septic embolus, providing further evidence of infection, and a potential cause of his seizures.
The patient’s sepsis was aggressively treated with intravenous hydration and broad-spectrum antibiotics, as well as ventilatory support. Blood cultures were sent. Given the evidence of infectious disease affecting both neurologic and cardiac systems, a presumptive diagnosis of infective endocarditis was made in the ED. This was confirmed on TTE.
While the incidence of infective endocarditis is somewhat difficult to determine given varied definitions of the disease, it is likely that the incidence ranges from 1.4 to 6.2 cases per 100,000 person-years.12 Infective endocarditis has increasingly become a disease of the elderly, with more than half of cases occurring in patients over the age of 60 years.13 This change is likely due to the fact that valve replacements occur most commonly in this age group and also because of decreasing incidence of rheumatic heart disease, which previously affected younger patients.
Infective endocarditis affects both native and prosthetic valves. The majority of cases of endocarditis are caused by staphylococci and streptococci, but numerous microbes have been identified as causative agents. Gram-positive organisms tend to predominate because they are more adherent to the endothelial surface of valve leaflets than Gram-negative organisms.14
The pathophysiology of endocarditis begins with turbulent flow adjacent to a heart valve, which can cause trauma to the endothelial surface of the heart. This, in turn, stimulates thrombus formation, which in the presence of bacteremia leads to the development of a bacteria-filled vegetation. Risk factors for infective endocarditis are directly related to the pathophysiology. Turbulent blood flow through the heart can be caused by conditions such as structural heart disease and prosthetic heart valves, while increased risk of bacteremia is present in those with a history of intravenous drug use and chronic hemodialysis. Other risk factors include a history of endocarditis, pregnancy, and colon cancer. Some patients, like the one presented in this case, have no recognizable risk factors for endocarditis.
The presentation of endocarditis usually includes constitutional complaints such as fever and chills, which can be nonspecific and make diagnosis difficult. Consequently, patients often present late in the course of illness with complications of the disease. The most common complications of endocarditis are cardiac, particularly heart failure and perivalvular abscess. As demonstrated in the case described above, acute coronary syndrome, while rare, may also complicate endocarditis.15 This may be due to either embolism or coronary compression from a periannular abscess. Debate exists surrounding the treatment of acute coronary syndrome associated with endocarditis. It is unclear whether septic emboli are as amenable to percutaneous coronary intervention as atherosclerotic thrombi are. Thrombolysis is not recommended due to the high prevalence of concomitant CNS disease and hemostatic abnormalities due to sepsis. In addition, bacteremia leads to an increased risk of intracerebral hemorrhage.16
Other common complications are related to septic embolization, such as pulmonary abscess, osteomyelitis, or renal infarction. Neurologic disease, including cerebrovascular accident, seizure, or abscess, may also be the result of systemic embolization. As in the case described above, up to 30% of patients with endocarditis present with focal neurologic complaints.17
The diagnosis of infective endocarditis in the ED requires a high clinical suspicion for the disease. As mentioned previously, the history can be vague, and risk factors may be absent. A heart murmur, traditionally taught as the key physical finding in endocarditis, is absent in about 15% of patients with endocarditis.18 Blood cultures should be ordered in the ED, although they will not aid in the initial diagnosis. Other laboratory studies are also unlikely to be useful in the diagnosis, although like the patient in this case, 40% of patients with endocarditis have hyponatremia.19
All patients in whom the diagnosis of endocarditis is being seriously considered should have echocardiography performed. Beyond serving as an essential diagnostic modality, echocardiography also provides an assessment of valvular dysfunction and identification of complications such as perivalvular abscess. TTE is often readily accessible and has a specificity close to 100%. It should be noted, however, that the sensitivity of TTE ranges from 29% to 63%.20 Transesophageal echocardiography (TEE) has a higher sensitivity (94–100%) than TTE and has improved utility in patients with prosthetic valves and in the detection of perivalvular abscess.20,21 Beginning diagnostic studies with TTE is recommended for most patients; however, in those patients for whom an especially high clinical suspicion for endocarditis exists, subsequent TEE may be required, even in the setting of a normal TTE.
Emergency department interventions center on disease identification, supportive care, and empiric antibiotics, ideally after blood cultures have been obtained. Febrile patients who are at high risk of endocarditis, such as patients with prosthetic valves or a history of intravenous drug use, should be admitted for further evaluation. Treatment for endocarditis includes antibiotics and, in some patients, valve replacement surgery. Despite these interventions, the mortality of endocarditis is approximately 25%, but varies based on patient characteristics, causative agent, and the development of complications.