Scrubbing Arteries for Stroke

Motivation: I made many resolutions this new year – one of them was to restart this blog. Many competing interests made me delinquent, but I am resolved to continue learning! For the first article of this year, we will turn to a long-standing issue in stroke neurology. Does intra-arterial therapy for stroke improve outcomes? Two years ago, I had thought about interventional neurology as a career but was dissuaded by the negative trials. Is intravenous tPA going away in 2015?

Paper: Berkhemer OA, Fransen PSS, Beumer D, et al. "A Randomized Trial of Intraarterial Treatment for Acute Ischemic Stroke." NEJM (2015); 372(1): 11-20.

Methods: A randomized, multi-center, open label but endpoint blinded trial in Netherlands in which adults admitted within 6 hours of stroke onset with imaging evidence of anterior arterial circulation occlusion were randomized to usual care (including IV tPA) or usual care plus intraarterial therapy (which consisted of mechanical thrombectomy and/or intraarterial thrombolysis). The primary outcome was modified Rankin scale at 90 days (0-6 with 2 or less indicating functional independence).

Results:
Cohort: 500 subjects (233 in intervention arm and 267 in control group) were randomized with mean age of 65 years (range 23 to 96) and 58% males. Median NIH stroke scale was 17 in intervention group and 18 in control. Around 80% in both groups had prestroke modified Rankin scale of 0. IV tPA was used in 87% and 91% of subjects in intervention and control group respectively. Most common imaging finding was M1 segment MCA artery occlusion. Median time from stroke onset to randomization was 204 and 196 minutes respectively in intervention and control groups. In the intervention group, time from stroke onset to groin puncture was 260 minutes.

Clinical Outcome: At 90 days, the median mRS was 3 in intervention group and 4 in control (unadjusted odds ratio of 1.66, 95% CI of 1.21-2.28). Adjusting for variables such as time from stroke to randomization, diabetes, atrial fibrillation, etc., the adjusted odds ration is 1.67 (95% CI: 1.21-2.30). When examining for independence (mRS of 0-2), 32.6% in intervention arm vs 19.1% in control group had mRS of 0-2 at 90 days (OR of 2.16, 95% CI 1.39 to 3.38).

Radiological Outcome: No intracranial occlusion was present in 75.4% in intervention group and 32.9% in control group. Final infarct volume was 49 mL in intervention group and 79 mL in control group.

Adverse Effects: No overall difference in serious adverse effects. However, embolization into new territory occurred in 8.6% of interventions, new dissection from manipulation in 1.7%, and vessel perforation in 0.9%.

Discussion:The trial shows that addition of intraarterial therapy to conventional IV tPA may have clinical benefit. This trial is in contrast to prior ones which did not show a benefit. Two reasons (among many) are that for part of the trial, the procedure was only offered within the context of the trial resulting in inclusion of many patients. In the US, many who are likely to benefit get the procedure anyway and are not included. Secondly, all the patients in this trial unlike previous ones had known arterial occlusion by imaging prior to randomization. While the results are encouraging, this new trial could also be the one lucky one showing benefit among many failed trials. Another major deficit was that the patients were not blinded and may have been biased while reporting! We will likely need follow-up corroborative trials elsewhere prior to accepting this therapy.

Opening Pressure in Children

Motivation: I thought that the magic number for lumbar puncture was 20 cm H2O - the upper limit of normal for opening pressure. I was recently measuring the opening pressure in a 10 year old with headache, and I wondered if the opening pressure limit is the same in children. After all, the space for the cerebrospinal fluid (CSF) space is differently shaped as well as the volume of the CSF present. Turns out amazingly that this issue was addressed only in 2010 in a New England Journal publication.

Paper:  Avery, RA, Shah, SS, Licht, DJ, et al. Reference Range for Cerebrospinal Fluid Opening Pressure in Children. NEJM (2010) 363: 891-893.

Methods: A 2 year prospective study of CSF opening pressure in children between ages 1 to 18 without signs or symptoms of raised intracranial pressure or diseases which might change the opening pressure such as meningitis. The study was conducted at Children's Hospital of Philadelphia.

Results: In total, 197 children met inclusion criteria. The 90th percentile for opening pressure was 28 cm H2O while the 10th percentile was 11.5 cm H2O. Age was not correlated with increased opening pressure. Variables correlating with increased pressure included moderate to deep sedation (p = 0.002) and higher BMI (weak correlation, coefficient = 0.313). For children who were not sedated, the 90th percentile for opening pressure was 25 cm H2O.

Discussion: Compared to adults, the opening pressure for children is likely higher, and when interpreting lumbar puncture results, mildly elevated opening pressure of 25 cm H2O should not be called abnormal. There are a few questions raised by these findings though. Why is the opening pressure in general higher for children? Is it from decreased circulation space or increased relative volume of CSF? Also, I do not understand why sedation, which in general is a relaxant, should increase the opening pressure rather than decrease it. Answers to these questions may be important in management of increased intracranial pressure in children.

The Sweet Intolerance

Motivation: The last few months have been a haze - the typical residency experience. Apologies for the sparse posts. We will begin again with a relatively shocking news topic. The reputation of artificial sweeteners is under attack with allegations that glucose tolerance levels may worsen with these artificial sweeteners. Is this true or craze?

Paper: Suez, J, Korem, T, Zeevi, D, et. al. "Artificial sweeteners induce glucose intolerance by altering the gut microbiota." Nature (2014) epub.

Methods: This study consisted of two parts. The first part was in mice divided into cohorts fed water mixed with saccharin, sucralose, aspartame, or glucose/sucrose.  The metabolic profile was then determined of these mice. The second part consisted of observational human data and a small trial of saccharin in control human adults.

Results:
Non-Caloric Sweeteners in Mice: After 11 weeks of mice being fed saccharin, sucralose, aspartame, or glucose/sucrose, all three artifical sweetener cohorts had significantly worse glucose tolerance with saccharin being the worst offender. Note that these doses of sweeteners where under the FDA approved dose limit by weight. The authors replicated this finding for saccharin in two other mice cohorts randomized to either glucose or saccharin.

The author's next postulated that this effect is mediated by the gut bacteria. Treatment with ciprofloxacin and metronidazole or vancomycin abolished the glucose intolerance effect of artificial sweeteners. Furthermore, fecal transplant from mice consuming saccharin also induced glucose intolerance in the transplanted mice. The authors further postulate that short chain fatty acid production is increased in the gut with artificial sweeteners and may be part of the causal chain to increased glucose resistance.

Non-Caloric Sweeteners in Human Beings: 381 non-diabetic adults (mean age 43.3) were studied next in a cross-sectional manner. Increased consumption of non-caloric sweeteners was associated with increased weight, higher fasting blood glucose, elevated glycosylated hemoglobin, and impaired glucose tolerance test.  These associations held true independent of adjustment for body mass index.

A small trial was conducted next in seven adults who do not consume artificial sweeteners. For six days, they were fed daily saccharin up to 5 mg/kg (FDA maximum limit). After this six-day experiment, 4 out of 7 subjects developed impaired glucose tolerance. Transplant of the feces from the subjects with impaired glucose tolerance into mice elicited impaired glucose tolerance in the mice suggesting that a change in the gut microbiome may be causal.

Discussion: This paper strongly shows that use of artificial sweeteners (particularly saccharin) is associated with impairment in glucose tolerance.  Although this relationship is shown most conclusively in mice, the small trial also suggests a similar process occurring in adult human beings. Furthermore the elegant fecal transplant experiments suggest that alterations in gut flora are probably to blame. This study raises serious concerns about the artificial sweeteners. However, prior to discarding the artifical sweeteners, we need larger scale observational human studies verifying that the imbalance in glucose tolerance seen within the short duration of the trial does in fact translate to increased risk of long-term glucose intolerance and development of diabetes. Moreover, these studies again highlight the fact that when these sweeteners were approved by the FDA, no clinical endpoints were used!

Renal Denervation for Hypertension

Motivation: Yes, the blog is not dead.  It has been moribund for a while under pressure from residency. Apologies.

A previous blog post here on July 2012 had considered kidney denervation for treatment of hypertension. Recently, there was a larger trial of this technique (SYMPLICITY HTN-3) was reported. So, should be all be considering this treatment for refractory hypertension?

Paper: Bhatt, D.L., Kandzari, D.E., O'Neill, W.W. et. al. "A Controlled Trial of Renal Denervation for Resistant Hypertension." NEJM (2014); 370: 1393-401

Method: Prospective blinded randomized sham-controlled trial in adult patients with severe resistant hypertension. Subjects were randomized in 2:1 ratio to undergo renal denervation or a sham procedure. Severe resistant hypertension was defined as three drugs at maximum doses. Primary endpoint was change in systolic blood pressure at six months.Of note, patients with renal artery stenosis were excluded.

Results:
Cohort: Total of 535 patients were randomized in this trial (364 to renal denervation group and 171 to sham procedure). The average age was about 57 years with roughly 60% of male patients. Baseline characteristics including body mass index, race, kidney disease, cardiovascular disease, and rate of smoking were not different between the two groups. On average, subjects took five antihypertensive medications.

Primary outcome: At six months, the intervention group had 14.13 mmHg decrease in SBP compared to 11.74 mmHg decrease in the sham group. This difference was not statistically significant.

Safety: There were five adverse effects in the intervention group compared to one in the sham group. There was no difference kidney function in the intervention group.

Discussion: This trial shows that renal denervation is not worth pursuing in the severely refractory hypertensive group of patients. The question remains why the therapy appeared so promising in the smaller trial but failed now.  One possible reason is that in the smaller previous trial, the placebo and the intervention groups were not well matched while the two groups are well matched in this trial. While this trial did not have promising results, one wonders whether this same procedure might benefit patients with early hypertension who may still respond to renal denervation. However, for now, I would not recommend this therapy.

Radiation Dementia Treatment

Motivation: Dementia is a dreadful word - a terminal cognitive decline ending in dependence and ultimately death.  Among the many causes, prior history of whole brain radiation is a risk factor.  Recently, a friend in radiation oncology told me that there was a trial for this kind of dementia with memantine.  How good is it?

Paper: Brown, PD, Pugh, S, Laack, NN et. al. "Memantine for the prevention of cognitive dysfunction in patients receiving whole-brain radiotherapy: a randomized, double-blind, placebo-controlled trial." Neuro-Oncol (2013); 15 (10): 1429-37.

Method: Eligible patients were adults with metastatic solid cancer to the brain receiving whole brain radiation with good performance status without renal failure and mini-mental status exam score > 18.  Patients were randomized to placebo or memantine (titrated up to 10 mg twice daily for length of 24 weeks).  Primary end-point was cognitive function (as measured by Hopkins Verbal Learning Test-Revised) at 24 weeks.

Results:
Cohort: Total of 508 patients randomized with 56% female.  Most common malignancy is lung cancer (70%).  256 were randomized to memantine while 252 were randomized to placebo. There were no differences in baseline age, gender, neurological functional status, education, prior history of radiation, or baseline neurocognitive score.

Compliance: Study compliance for the drug was 31% for memantine arm and 33% for placebo arm.  The most common reasons for discontinuation were patient refusal and patient death followed by disease progression or adverse event (similar in both groups).

Outcome: Compared to the placebo group, there was decreased decline in Hopkins Verbal Learning Test-Revised scale (0 for memantine vs. -0.9 for placebo, p = 0.059).  There was a statistical benefit in terms of Mini Mental Status Exam (0 vs -1, p = 0.0093).  There were no differences in progression-free survival or overall survival.

Adverse Effect: No statistical difference in adverse effect between groups.

Discussion: This trial seeks to hint that memantine may have some cognitive benefit in whole brain radiation, but the trial is seriously hampered by the very poor study compliance (2/3 or more of the patients did not comply with study assignment).  Also, the statistically significant Mini-Mental Status Exam improvement by one point is likely not clinically meaningful.  At present, I would say that the efficacy of memantine remains unanswered awaiting further testing.

Pain in Sjogren's Syndrome

Motivation: Every time someone comes with painful feet, I ask "Do you have dry mouth or dry eyes?" Patients usually look at me suspiciously.  What does dry eyes have to do with painful feet?  Sjogren's syndrome, of course  - an elusive syndrome of dry eyes, dry mouth, and autoimmune destruction of exocrine glands often with associated peripheral neuropathy.  To avoid those suspicious glances, I wonder what are the characteristics of neuropathy from Sjogren's syndrome.

Paper: Berkowitz, AL and Samuels MA. "The neurology of Sjogren's syndrome and the rheumatology of peripheral neuropathy and myelitis." Pract Neurol (2013); 0: 1-9.

Methods: Review of peripheral nervous system presentations of Sjogren's syndrome and characteristics of serologic testing.  This is part of a broader paper reviewing Sjogren's associated myelitis as well.

Results:
Prevalence: Neuropathy accompanies Sjogren's syndrome in approximately 5-15% of cases.  Neuropathy preceded other symptoms in 37%, occurred concurrently in 16%,and occurred after other symptoms in 37%.

Types of Neuropathy: The most common forms of neuropathy involve the dorsal root ganglia (hence pure sensory loss) in 39% and small unmyelinated fibers (hence painful) in 20%.  Other presentations included trigeminal neuropathy in 16%, multiple mononeuropathies in 12%, multiple cranial neuropathies in 5%, and polyradiculoneuropathies in 4%.

Testing: The classical serum auto-antibodies anti-Ro (SSA) and/or anti-La (SSB) occur in 10-55% of patients with Sjogren's neuropathy.  For dorsal root involvement, the sensitivities of SSA and SSB are 53% and 11%.  For painful small fiber neuropathy, the sensitivities of SSA and SSB are 39% and 17%.  Anti-nuclear antibody (ANA) is positive in 20-67%.  Schirmer's test evaluating tear production (degree of moistening of filter paper in lower eyelid after 5 minutes) is positive in 56-89%.  Lip salivary gland showing lymphocytic infiltration is diagnostic in 37-75% of patients.

Discussion: 
The first depressing conclusion of this paper is that the manifestations of Sjogren's syndrome are protean without good diagnostic tests.  Nonetheless, Sjogren's associated neuropathy is most often sensory in nature (dorsal root gagnlia or just painful neuropathy) without significant motor involvement.  The second point is that serologic testing may be negative in about 50% or even more number of patients.  Similarly, classical symptoms of dry eyes or dry mouth may follow neuropathy and should not be used to exclude Sjogren's syndrome.  In the office, stocking filter paper and learning to perform the Schirmer's test may be more helpful than sending for serologic testing.  

Lipoprotein (a) Value

Motivation: Lipoprotein (a) inspires strong emotions among doctors.  Some like it as a risk factor while others see it as a waste of money.  But, when controlled for conventional lipid markers, is it indicative for additional risk for cardiovascular disease?  For background, lipoprotein (a) is a low density LDL-like particle synthesized by the liver which is found in the intima of arteries and presumably promotes atherosclerosis.

Paper: The Emerging Risk Factors Collaboration. "Lipoprotein (a) Concentration and the Risk of Coronary Heart Disease, Stroke, and Nonvascular Mortality."   JAMA (2009); 302: 412-423.

Methods: Meta-analysis of long-term prospective studies that recorded Lipoprotein (a) (Lp(a)) and vascular morbidity.

Results:
Studies: 36 prospective studies met inclusion criteria.  In the analysis, 126 634 participants were included for 1.3 million person-year of follow-up with 22 076 vascular disease outcomes or death.  Mean age at entry was 57 years with 48% women.  Ethnicity was 47% European and 50% North American.

Lipoprotein (a): At baseline, the overall population Lp (a) was 12.6 mg/dL.  Blacks had 119% (95% CI: 84 to 161) higher Lp(a) concentration compared to whites at baseline.  Women had 12% (CI: 8 to 16) higher Lp (a) than men.

Coronary Heart Disease (CHD): When adjusted for age, sex, systolic blood pressure, smoking, diabetes, and total cholesterol, the relative risk of coronary heart disease for the top third of Lp (a) compared to the bottom third was 1.27 (95% CI: 1.17-1.38).  In the top third of subjects with Lp (a), the rate of CHD was 5.6 (95% CI: 5.4-5.9) per 1000 person years compared to rate of 4.4 (95% CI: 4.2-4.6) per 1000 person years in the bottom third.

Ischemic Stroke: When adjusted for usual risk factors, the relative risk (RR) for ischemic stroke was 1.10 (95% CI: 1.02-1.18) per 3.5 fold higher than usual Lp (a) levels.  The relative risk did not reach significance for unclassified stroke and hemorrhagic stroke.

Non-vascular mortality: Lp (a) levels were not associated with increase in non-vascular mortality.

Discussion: Lipoprotein (a) is very modestly associated with independent risk of coronary heart disease and ischemic stroke.  As stated in the paper, compared to the power of Lp (a), elevated non-HDL cholesterol level is four times more strongly associated with coronary heart disease.  There is no drug that independently targets Lp (a) levels.  Consequently, I do not think that at present, it is worth measuring this modestly predictive marker without clear treatment.  The paper, though, illuminated the variability of Lp (a) across ethnicities (blacks have baseline of 100% greater than whites) and genders.  When interpreting raw values of Lp (a), we have to be careful about using correct gender and ethnicity matched norms, which may not exist in all cases.

Cooling Dogs

Motivation: Lately, hypothermia has not been good enough for me.  There have been too many people who have suffered cardiac arrests at home and ended up severely impaired despite receiving mild hypothermia according to protocol.  Can we do better?  For instance, if we can quickly institute deep hypothermia, is it any better?

Some of the coldest studies in this field have been performed in dogs.  In this amazing study, dogs were tested for neurological recovery after many hours of hypothermia.

Paper: Nozari A, Safar P, Wu X, et. al. "Suspended Animation Can Allow Survival without Brain Damage after Traumatic Exsanguination Cardiac Arrest of 60 Minutes in Dogs." J Trauma (2004); 57: 1266-1275.

Methods: Fourteen dogs were sedated with ketamine and halothane and supported with positive pressure ventilation.  The dogs were then exsanguinated over fiver minutes to cardiac arrest.  At two minutes, flush of saline at 2 degrees Celsius was administered in femoral artery to achieve tympanic temperature of 10 degrees Celsius.  In 6 dogs, reperfusion was performed after 60 minutes of cardiac arrest.  In 8 dogs, to simulate trauma, splenic injury was inflicted prior to cardiac arrest.  During arrest time of 60 minutes, the spleen was transected.  Outcome was determined by overall performance category (1 to 6 with 1 being normal and 6 being death) at 72 hours.

Results:
Resuscitation: All fourteen dogs were successfully resuscitated from cardiac arrest.

Control group: All six dogs in the control group were neurologically normal after 72 hours (they were judged to be identical in behavior to their pre-intervention condition).

Trauma group: Of the eight dogs who underwent splenectomy during the cardiac arrest time, four were normal after resuscitation.  One had moderate disability while another dog had severe disability.  Two dogs remained in coma.

Discussion: In this remarkable study, dogs could be maintained in suspended animation at 10 degrees Celsius when rapidly cooled after two minutes.  The most obvious problem in extending this strategy to the real world is that Emergency Medical response time is certainly greater than two minutes.  However, while we know that delayed cooling leads to increased brain injury, trials using suspended animation have not been tried in human beings even in settings in which prompt response is possible (such as hypothermia instituted in the field).  One of the other interesting findings in the study is that operative trauma during suspended animation (splenectomy) leads to a worse outcome after reperfusion.  The cause is unclear in this case but could be secondary to need for increased recovery times (all dogs were timed at 72 hours and not at best long term performance) or possibly injury from increased systemic inflammation.

Proton pump inhibitors in liver cirrhosis

Motivation: A patient with known liver cirrhosis is admitted. His home medication list includes a proton pump inhibitor (PPI). If there is no other indication (such as peptic ulcer disease), what is the evidence for use of PPIs in liver cirrhosis - in particular, are PPIs effective in preventing GI bleed in liver cirrhosis? Are there any risks associated with PPI use?

Literature:
1) LEVEL OF EVIDENCE: retrospective study
 Garcia-Saenz-de-Sicilia M et al. PPIs are not associated with a lower incidence of portal-hypertension-related bleeding in cirrhosis. World J Gastroenterol. 2010 16(46):5869-73.

This was a restrospective study that included 105 patients with cirrhosis with endoscopy-proven portal hypertension. Patients were divided into 2 groups: those who used PPIs (45.5%) and those who did not. Seventeen (16.1%) patients had upper GI bleeding, due to either esophageal varices or portal HTN gastropathy. Of these 17 patients 9 used PPIs, while 8 did not use PPIs (p=0.51). Thus, in this retrospective study, use of PPIs was not associated with development of GI bleed related to portal HTN.

2) LEVEL OF EVIDENCE: literature review

Siple JF et al. Proton pump inhibitor use and association with spontaneous bacterial peritonitis in patients with cirrhosis and ascites. Ann Pharmacother. 2012 46(10):1413-8.

This was a literature review that evaluated the use of PPIs in cirrhosis and ascites, ultimately including 1 case study, 8 restrospective case-controlled studies, 1 meta-analysis. Authors summarize that there is a possible risk of association between use of PPIs and increased risk of SBP and C diff infection in cirrhosis/ascites. However, they note that this association has yet to be established by prospective trials.

3) LEVEL OF EVIDENCE: RCT 

Shaheen NJ et al. Pantoprazole reduces the size of postbanding ulcers after variceal band ligation: a randomized, controlled trial. Hepatology. 2005; 41(3):588-94

In this study, the use of PPI after elective esophageal variceal ligation (EVL) was investigated. This study was designed as a double-blinded, placebo-controlled RCT where 44 subjects (42 completed the study) who underwent EVL were randomized to the pantoprazole arm (pantoprazole 40 mg IV after EVL x 1 then pantoprazole 40 mg PO x 9 days) or the placebo arm (placebo IV x1 after EVL then placbo PO x 9 days). Upper GI scope 10-14 days after the EVL procedure showed that while both groups had the same mean number of ulcers, ulcers in the pantoprazole arm had smaller average size (p < 0.01). 

Thoughts: I was unable in my initial literature search here to find definitive evidence that PPIs should be used for primary prevention for GI bleeding secondary to portal hypertension in liver cirrhosis. One retrospective study by Garcia-Saenz-de-Sicilia et al. (2010) did not show differences between patients who took PPIs and those who did not, but this was a relatively small sample that was studied retropectively. The RCT by Shaheen et al. (2005) provide evidence for using pantoprazole directly after EVL, though it was a short, limited course of PPI use that was studied. Authors Siple et al. (2012) recently draw from existing literature to caution on the potential risk of infection (SBP, C diff) when using PPIs, though there is no prospective, RCT evidence for this concern. It is important to think about the indications of using PPIs in patients, realizing that there lacks evidence for use of PPIs in primary prevention of GI bleeds related to portal HTN in liver cirrhosis (Garcia Saenz-de-Sicilia et al., 2010). While one should be aware of possible adverse effects like increased risk of infection (Siple et al., 2012), there also does not exist evidence against using PPIs in liver cirrhosis if there is a reasonable primary indication for PPI use (e.g. peptic ulcer disease).

Blood Pressure Goals

Motivation: 150?? Since medical school, we have been firmly taught that the upper limit of normal for systolic blood pressure is 140 mmHg.  Recently, the panel JNC8 ruled that for adults over 60, a reasonable systolic goal is less than 150 mmHg.  After having counseled countless times about the importance of blood pressure control, this increase sounded traitorous.  What is the data?  There have been two randomized trials addressing this issue over the past decade.  I will review here the one of the larger and latest one here.

Paper: Ogihara, T, Saruta, T, Raguki, H, et. al.  "Target Blood Pressure for Treatment of Isolated Systolic Hypertension in the Elderly" Hypertension (2010) 56: 196-202.

Methods: The study (VALISH) was a multicenter, prospective, randomized, open-label, blinded end point trial in Japanese adults between 70-85 years of age with isolated systolic hypertension (SBP > 160 mmHg with diastolic less than 90 mmHg).  Patients were randomly treated with valsartan (titrated first with addition of second agent if necessary) into two groups: (1) Systolic blood pressure < 140 mmHg (strict control) or (2) systolic blood pressure < 150 mm Hg (moderate control).  Patients were followed for minimum of two years.  Primary outcome was a composite of cardiovascular events (sudden death, stroke, MI,  death from cardiovascular cause, renal dysfunction).

Results:
Cohort: A total of 3260 patients were randomized.  There were 181 patients lost to follow-up (95% follow-up).  In total, 3079 patients were followed for average of 2.85 years.  Average age was 76.1 years with 62% women.  Baseline characteristics were mostly balanced between groups except there were more smokers in the strict control group vs. moderate control (21% vs. 17.4%; p = 0.01).  There was history of stroke in 6.5%, ischemic heart disease in 5%, heart failure in 1.7%, and diabetes mellitus in 13%.

Blood Pressure Control: At 36 months of follow-up, the blood pressure was 13.6./74.8 in strict control group and 142/76.5 in moderate control groups.  Heart rate did not differ significantly between the two groups.

Outcome: In the primary combined cardiovascular outcome, the overall rate did not differ between the groups in intention-to-treat analysis (10.6 events per 1000 patient years in strict control; 12 per 1000 patient years in moderate control group, p = 0.38).  There was no difference in any of the individual components of the composite outcome.  In subgroup analysis, there was not a statistical difference in patients with diabetes, dyslipidemia, and chronic kidney disease.

Adverse Effect: There was no difference in adverse effect between the two groups.

Discussion: After an average of 2.8 years of follow-up, there did not appear to be a difference in cardiovascular outcome by controlling blood pressure tighter in patients with isolated systolic hypertension.  While rather remarkable and against popular medical conception, I think that the lack of longitudinal data beyond 2.8 years is cautionary is depending on these blood pressure parameters too much.  This trial shows that systolic blood pressure control less than 150 mmHg versus 140 mm Hg may not make much of a difference in the short term, but what about in five to ten years?  We still do not know the answer.  Another problem with extending this dataset is that patients included in the cohort had low burden of disease (ischemic heart disease in only 5%).  For the patient with angina, the trial may not be powered enough to detect difference in this subgroup.    But then again, these arguments are probably my ingrained cognitive biases against change.  The bottom line, I think, is that we need more longitudinal data before having a blood pressure parameter in mid.

Stroke Recovery for How Long

Motivation: So, how long does it take to recover? Even after more than a year of treating patients with acute stroke, I am not sure how to answer this question.  I sometimes put it vaguely as "months."  But, really, after how many months do most post-stroke patients complete their recovery?

Paper: Jorgensen HS, Nakayama H, Raaschou HO, Vive-Larsen J, Stoier M, Olsen TS. "Outcome and time course of recovery in stroke: Part II: Time course of recovery. The copenhagen stroke study." Arch Phys Med Rehabil (1995); 76: 406-412.

Methods: All patients with acute stroke in Copenhagen, Denmark between September, 1991 to 1993 were followed from time of acute admission to end of rehabilitation to six months post-stroke.  Time course of recovery was plotted.

Results:
Cohort: The cohort of stroke survivors consisted of 947 patients (53% female) of mean age 73.3.  The strokes were 93% ischemic and 7% hemorrhagic.  The median time between symptom onset and admission was 16 hours.  After rehabilitation, 19% were placed in nursing homes while 81% were discharged home.

Recovery: When assessed by the Scandinavian Stroke Scale (0-58 points, higher indicating milder deficits), best neurological recovery was reached in 80% by 4.5 weeks (95% CI: 4 to 5 weeks) and in 95% by 11 weeks (95% CI: 10.1 to 11.9 weeks) from stroke onset.  Best ADL function (measured by Barthel Index) was reached in 80% by 6 weeks (95% CI: 5.3 to 6.7 weeks) and in 95% by 12.5 weeks (95% CI: 11.6 to 13.4).

Initial Severe Functional Deficits: Severity of functional deficits were judged by the Barthel Index (scale of 0 to 100 with very severe disability in 0-20 and no disability with score of 100).  In those with very severe initial disability (index of 0-20), best ADL function was reached in 80% of patients within 11 weeks (95% CI: 10-12) and in 95% within 17 weeks (95% CI: 15-19).

Initially Mild Disability: In those with mild disability (Barthel Index 75-95), best ADL function was reached in 80% of patients within 2.5 weeks (95% CI: 2-3) in 95% of patients within 5 weeks (95% CI: 4-6).  

Discussion: In this remarkable study with 100% follow-up of all stroke patients in Copenhagen over two years, the overall message is that 95% will regain their best ADL function in about three months.  The rate of recovery is slower in those with more severe disability and more rapid in those with mild disability.  However, even in those with very severe initial functional deficits, best ADL function was reached in 17 months.  For general counseling purposes, an appropriate summary statement might be that mild strokes take about a month to regain best function while severe strokes take about four to five months to regain best function.  While this study is well done, some cautionary aspects are that (1) acute stroke therapy has changed since  the study, and (2) physical therapy has changed since the early 1990s.  How these aspects change the natural history of stroke disorders is unclear.  Also note that this study also talks about the rate of recovery and not the extent of recovery.

Time to Defibrillation and Survival - How soon?

Motivation: Last year, a man walked into my clinic, shook my hand, and said that he had survived a v-fib cardiac arrest.  My heart skipped a beat.  He told me that he had been unconscious for more than ten minutes.  I did not know whether to believe him.  What is the relation between survival and time to defibrillation in out of hospital cardiac arrest?

Paper: De Maio, V.J., Stiell, I.G., Wells, G.A. et al. "Optimal Defibrillation Response Intervals for Maximum Out-of-Hospital Cardiac Arrest Survival Rates"  Ann Emerg Med. (2003) 42: 242-250.

Methods: Prospective cohort study of the Ontario Prehospital Advanced Life Supports (OPALS) study, in which 21 Ontario study communities received a basic life support level of care with defibrillation by EMS.  The study assessed the relation of survival to defibrillation response time by EMS between 1991-1997.

Results:

Cohort: There were a total of 9,273 out of hospital cardiac arrests.  The mean age of 68.3 with 67.7% male sex.  The initial rhythm was ventricular fibrillation in 38.5%, asystole in 42.3%, and pulseless electrical activity in 19.1%.  There was return of spontaneous circulation in 10.4%.  Of the total 9,273 patients with cardiac arrest, 392 were discharged alive from hospital (4.2%).

Defibrillation: Of the 9,273 cardiac arrests, 4,059 received defibrillation (43.8%).  The median time to defibrillation was 6 minutes.  The 90th percentile for defibrillation response was 9.3%.

Survival:  When survival to hospital discharge is related to response time to defibrillation:

• Response time less than 4 minute: 7.6% survival.
• Response time between 4-6 minutes: 4.0% survival
• Response time between 6-8 minutes: 2.8% survival
• Response time greater than 8 minutes: 1.6% survival

When survival is modeled continuously to time to defibrillation, the predicted survival is (90th percentile prediction):

• Defibrillation time 1 min: 28.1% survival
• Defibrillation time 2 mins: 23.1% survival
• Defibrillation time 3 mins: 18.8% survival
• Defibrillation time 4 mins: 15.1% survival
• Defibrillation time 5 mins: 12.0% survival
• Defibrillation time 6 mins: 9.5% survival
• Defibrillation time 7 mins: 7.5% survival
• Defibrillation time 8 mins: 5.9% survival

Discussion: I think that the overall lesson is that surviving an out of hospital cardiac arrest is hard!  Even with good response time between 4-6 minutes, there is only a 4% survival.  While defibrillation definitely appears to increase survival (by almost four fold when comparing response time less than 4 minutes to greater than 8 minutes), the really steep drop in survival comes at the very beginning of the cardiac arrest (for example, almost 20% drop in predicted survival from two minutes to three minutes).  This study, however, has some flaws in accounting for the benefit of defibrillation.  Most notably, there were more defibrillations performed than there were ventricular fibrillations.  Consequently, some PEA and asystole arrests were also defibrillated without a clear indication.  Including these in the accounting probably dilutes the effect of defibrillation. 

Heart Failure Hospitalization - How Bad?

Motivation: Last year I saw an 85 year old man with a stroke.  After never having walked into a hospital or doctor's office, he was in the hospital for diastolic heart failure and was then found to have difficulty moving his left hand.  When talking to him about preventing future strokes and the necessary tests, I wondered whether his heart or his brain was his bigger problem.  What is the prognosis after hospitalization for heart failure?

Paper: Jong, P, Vowinckel, E, Liu, P et. al. "Prognosis and Determinants of Survival in Patients Newly Hospitalized for Heart Failure" Arch Intern Med (2002); 162 (15): 1689-1694.

Methods: Retrospective population-based cohort of 38,702 consecutive patients with first admission for heart failure in Ontario, Canada between 1994 and 1997.  Outcome was mortality at thirty days and at one year.

Results:
Cohort: The cohort of 38702 patients consisted of 51.1% women with 84.6% older than age 65 and 57.9% older than age 75.

Mortality: The net mortality rate was 11.6% at thirty days and 33.1% at one year.  Men had a slightly higher mortality rate at thirty days (OR: 1.09, p = 0.001) and at one year (OR: 1.16, p < 0.001).

Age Relation: Age very significantly affected the prognosis.  The one year mortality rate was 13.5% in those younger than age 50 and 40.1% in those 75 or older.

Comorbidities: Using the Charlson composite index (a composite scale with higher score meaning more number and severity of comorbidities), a patient with score of 0 had one year mortality of 26.8% while a patient with score of 3 or greater had mortality of 50.6%.  Among specific conditions, malignancy, renal disease, dementia, cerebrovascular disease, and previous myocardial infarction all independently and significantly increased one year mortality.

Conclusion: Excepting in those younger than fifty without significant comorbidities, the prognosis of hospitalization for heart failure is quite grim.  I think that a general figure to keep in mind is that roughly a third will be dead after one year.  For someone, with history of stroke, diabetes, and chronic kidney disease (a quite common combination with Charlson score of 3), roughly half will be dead in a year.  I do not think that families or patients are quite aware or counseled adequately of the implications of heart failure.  Of course, this study has some serious limitations as well - most particularly, therapeutics for heart failure have become more standardized with incorporation of beta-blockers, ACE inhibitors, and ischemic heart disease treatment than in the 1990s.  We really need a similar study on a more modern population though I am not sure that the conclusions will be that significantly different.

The anti-nuclear antibody

Motivation: Midway last year, the anti-nuclear antibody (ANA) test started appearing rather useless to me.  The ANA was positive at some titer in just about everyone we tested, and even when it was positive, we discarded it under the mantra of "non-specific."  What is the range of ANA in the "healthy" population?

Paper: Tan, E.M., Feltkamp, T.E.W., Smolen, J.S. et. al. "Range of antinuclear antibodies in "healthy" individuals." Arthritis & Rheumatism (1997); 40: 1601-1611.

Methods:   Fifteen international laboratories experienced in performing the ANA assay were asked to give samples from healthy individuals of different age groups and from patients with systemic lupus erythematosus (SLE), scleroderma, Sjogren's syndrome, rheumatoid arthritis (RA), and soft tissue rheumatism.  Healthy was defined as working individuals between 20-60 years of age without physical or mental disabilities.  Soft tissue rheumatism consisted of patients with non-articular pain.

Results:
Measurement Variability: Among the fifteen laboratories, inter-laboratory coefficient of variation for ANA was: 50.7% for 1:40 dilution, 44.3% for 1:80, 37.9% for 1:160, and 36% for 1:320.  Intra-laboratory variation was about three times lower than the inter-laboratory coefficient of variation.

Normal Individuals: Overall, 46.7% tested negative on ANA.  31.7% tested positive at 1:40 dilution, 13.3% at 1:80 , 5% at 1:160, and 3.3% at 1:320 dilution.

1:160 dilution: At the the 1:160 dilution, the ANA was 95% sensitive for SLE, 86.5% sensitive for scleroderma, 74% sensitive for Sjogren's, 13.5% sensitive for RA, and 7.7% sensitive for soft tissue rheumatism.  Compared to healthy controls, cutoff of 1:160 is 95% specific.

1:320 dilution: At the the 1:320 dilution, the ANA was 86.8% sensitive for SLE, 83.8% sensitive for scleroderma, 71.1% sensitive for Sjogren's, 2.7% sensitive for RA, and 3.8% sensitive for soft tissue rheumatism.  Compared to healthy controls, cutoff of 1:320 is 96.7% specific.

Discussion: A cutoff of 1:160 appears to me a more reasonable value to use to separate possible pathology from variation in healthy individuals with only 5% of healthy population having titers of ANA higher than 1:160.  To me, other surprising findings were how variable ANA is between laboratories and how poorly the ANA performs in detecting rheumatoid arthritis (only 13.5% sensitive for RA) even though it is a immunologic disorder with positive serologic markers.  These figures must, of course, be interpreted with caution because the healthy population was derived from random samples submitted to laboratories.  Aside from individuals having no physical disability, we do not know more specifics which could affect the performance of ANA.  For instance, would a population of young women or older individuals with cancer have different characteristics in the ANA?

Fever and Pulmonary Embolism

Motivation: Clinical lore has it that pulmonary embolism (PE) causes low grade fevers and should not cause a fever as high as 103F.  While deciding on the often pondered question of whether or not a person has PE, I have seen the degree of fever sway the decision one way or the other.  But, is this true?

Paper: Stein, PD, Afzal, A, Henry, J, et. al. "Fever in Acute Pulmonary Embolism" Chest (2000); 117:39-42.

Methods: In the PIOPED (Prospective Investigation of Pulmonary Embolism Diagnosis) study, retrospective analysis was made of presence of fever in patients with angiographically proven pulmonary embolism (PE).

Results:
Cohort: There were 363 patients from PIOPED study analyzed for this study.  Fever was present in 95 patients (26%; 95% CI: 22-31%).  Of the 95 patients with fever, 43 (45.3%) did not have other causes of fever - solely attributed to PE.  There were 46 (48.4%) patients who had an alternative possible etiology of fever - 15 with malignancy, 10 with infection, 6 with recent operation (top three causes).   Six did not have further investigation (so indeterminate cause).  Fever was not more frequent in patients with pulmonary hemorrhage or infarction than in those without.

Maximum Temperature: Of the 43 patients with available information on temperature, the maximum temperature was 100 to 100.9F in 24, 101 to 101.9F in 14, 102-102.9 in 3.  One patient each had maximum temperature between 103-103.9 and greater than 104F.

Discussion: The clinical lore turns out to be mostly true - majority of patients with PE associated fever have maximum temperature less than 102F.  What I thought was most cautionary, however, is that fever is an infrequent finding in PE (present in only about a fourth).  Even among those with fever, they were just as likely to have an alternate explanation as pulmonary embolism.  What I took away from this paper is that if patients with PE have fever, they still need a thoughtful workup.

Taking a Chance on Chantix

Motivation: Last month, I met a man in clinic who said that his life had changed after his stroke.  Now, after years of fruitless contemplation, he was going to quit smoking.  He asked me about Chantix (generic varenicline).  Is it safe?

My instinct was to say yes.  But, the partial nicotinic agonist effects also make varenicline a sympathomimetic agent.  In patients with cardiovascular disease, are there excess side effects?  Turns out that the issue has been tested in a randomized fashion in 2010.

Paper: Rigotti, N.A., Pipe, A.L., Benowitz, N.L., et. al. "Efficacy and Safety of Varenicline for Smoking Cessation in Patients with Cardiovascular Disease." Circulation (2010); 121: 221-229.

Methods: Multi-center randomized placebo-controlled blinded trial testing varenicline (1 mg daily; quit day eight days after start of therapy) versus placebo in adult (35 to 75 years) smokers with cardiovascular disease (coronary artery disease, peripheral vascular disease, or cerebrovascular disease) existing for more than two months.  Patients with depression, severe neurological deficits from stroke, or severe renal and hepatic diseases were excluded.  Treatment group received 12 weeks of therapy.  Primary outcome was carbon monoxide confirmed abstinence between weeks 9 to 12 after start of therapy.  Patients were followed for 52 weeks.

Results:
Subjects: In total, 714 subjects were randomized to varenicline (355) or to placebo (359).  Study completion rate was 82.8% in varenicline group and 85.1% in placebo group.  Baseline characteristics were comparable between groups.  In the varenicline group, average age was 57 with 75% male and 80% white.  Average smoking history was for forty years.  For CVD, majority had CAD.  Only 4.5% had suffered a stroke.

Efficacy: The abstinence rate at weeks 9 through 12 was 47% in the varenicline group versus 13.9% in placebo (p<0.0001).  The rate of abstinence from week 9 to end of follow-up at week 52 was 19.2% for varenicline versus 7.2% for placebo (p<0.0001).

Adverse Effect: Patients on varenicline suffered more adverse effects than placebo (81.6 vs. 64.9%, CI for difference: 10.3-23.2%) resulting in more participants who stopped drug in varenicline group (9.6%) versus placebo (4.3%).  There were no differences in mortality.  The most common side effect was nausea (29.5% in varenicline vs. 8.6%).  Sleep disorder consisting of abnormal dreams, insomnia, and nightmare was significantly more frequent in treatment group (22.1%) vs placebo (9.7%).  There was a trend towards higher frequency of cardiovascular events in varenicline group (7.1% vs 5.7%) which did not reach significance.

Discussion: In summary, varenicline does not appear to increase risk of cardiovascular events significantly within the power of the study.  It was sobering to see though that after a year, only 19% will be abstinent even after being motivated enough to enroll in a trial and take a drug for twelve weeks.  Varenicline is certainly not free of adverse effects with nausea (30%) and sleep disorder (22%).  Prior to prescribing, patients need to be warned of these potentially bothersome side effects.  Also, of note, the trial was conducted in patients with stable CVD (defined as no events within the past two months).  So, I would likely wait for two to three months before prescribing varenicline to the patient seeking to stop smoking.

Causes of Flapping

Motivation: Look - the hands are flapping.  One of the useful tests in a confused patients is to ask them to stop traffic with outstretched hands.  If the hands start flapping in a non-rhythmic way ('asterixis'), that is interpreted as proof of generalized toxic-metabolic encephalopathy.  But, then again, there is probably a neural pathway underlying this mechanism.  We often dismiss asterixis in a blase way, but can there be focal brain lesions underlying asterixis?

Paper: Degos, J-D., Verroust, J., Bouchareine, A., et. al. "Asterixis in Focal Brain Lesions"  Arch Neurol. (1979); 36: 705-707

Methods: At Henri Mondor Hospital (Creteil, France), one of the authors documented twenty cases of asterixis with focal brain lesions.  EMG was available for nine patients.

Results:
Midbrain Lesions: Four patients ranging from 38 to 62 years presenting with unilateral asterixis after abrupt onset weakness in the affected side.

Parietal Lesions: Seven cases of asterixis with pathologic proof in four occurred from parietal lesions.  These lesions could occurred on contralateral side from the asterixis and could be from mass lesions (glioblastoma, abscess, metastatic lesion) or from stroke (hemorrhagic or ischemic).

Suspected Midbrain and Parietal Lesions: Six cases with suspected lesions.  One of these cases was a right temporal glioblastoma invading the lenticular nuclei with left sided asterixis and another had an expanding right thalamic tumor with left sided asterixis.

Unclear lesions: Three cases of unilateral asterixis with accompanying unilateral weakness or numbness of unclear etiology.

EMG Recordings: During episodes of asterixis, there were 50-100 ms periods of electrical silence in both agonist and antagonist.

Discussion: This old paper shows that true asterixis is not purely a toxic-metabolic phenomenon.  Apparently, parietal, midbrain, and thalamic lesions can result in contralateral asterixis.  While this paper certainly does not suggest a pathophysiological explanation of asterixis, some important points are that: (1) asterixis appeared contralateral to the lesion meaning that usually the side with asterixis is the one that is weak or numb, (2) both cortical and deep lesions can result in asterixis,.  I think that this account provides a cautionary note of dismissing asterixis as purely toxic phenomenon.

Interfering with RNA

Motivation: In medical school, while watching cartoons depicting disease pathophysiology pathways, I have thought many times that if we could just somehow block that damaged protein, we will cure that disease.  In mice, of course, multiple techniques exist including creating mutant breeds, transfecting with viruses, or injecting short RNA interfering sequences.  I was thrilled to read last month that such magic has for the first time been tried successfully in human beings.  Given the momentous significance of the recent trial testing RNA interference in transthyretin amyloidosis, we will cover the technique and consequences.

Paper: Coelho, T., Adams, D., Silva, A., et. al. "Safety and Efficacy of RNAi Therapy for Transthyretin Amyloidosis" NEJM (2013); 369: 819-29.

Methods: This Phase I study consisted of two parts.  In the first portion, eligible patients were older than 18 years with biopsy proven transthyretin amyloidosis with mild-to-moderate neuropathy.  Patients were excluded if they had liver, renal, or thyroid dysfunction.  These patients were injected in a blinded way with lipid nanoparticles containing RNA interfering (RNAi) sequences or with placebo.  In the second part, seventeen healthy patients were injected with a second generation lipid nanoparticle containing RNAi sequences testing for safety.

The lipid nanoparticles consisted of ionizable lipid components with RNA sequence that targets a conserved sequence at the 3' untranslated region of mRNA of Transthyretin protein (TTR).

Results:
Cohort: In the cohort with disease, there were 32 patients with 56% male and median age of 37.  Majority of participants were recruited from Portugal (23/32 patients).  The healthy volunteer cohort were all 17 male subjects with mean age of 27 (all were from England).

Amyloidosis Cohort: Patients with TTR amyloidosis treated with placebo did not have change in serum transthyretin levels.  In the group receiving 1mg/kg lipid nanoparticle dose, there was a mean reduction of 38% in serum transthyretin levels with recovery to close to baseline levels by day 28.

Healthy Volunteers: Using the second generation lipid nanoparticles, peak reduction of serum transthyretin level was reached at approximately day 10.  At 0.3 mg/kg and 0.5 mg/kg doses, the mean levels of suppression were 82.3% and 86.8%.  The serum transthyretin levels remained suppressed even at day 28.

Safety: There were no drug related serious adverse events or study drug discontinuation.  No significant changes in hematologic, liver, or renal measurements were seen.  Mild to moderate infusion reactions occurred in 7 to 20% of reactions responding to temporary interruption of infusion or administration of glucocorticosteroids.  Antibodies to pegylated lipid components were not seen.

Discussion: This trial demonstrated successful use of lipid nanoparticle RNAi to suppress transthyretin levels.   In the four week follow-up, there were remarkably no drug related side effects.  This Phase I trial, while promising, does not show efficacy of treatment on important clinical variables.  Transthyretin is primarily synthesized by the liver but is also synthesized to a smaller extent in the retina and choroid plexus.  The hepatic suppression may still leave the patient vulnerable to retinal and CNS side-effects.  In normal physiology, transthyretin is associated with retinol binding protein levels and Vitamin A levels.  Long term suppression of transthyretin may also interfere with Vitamin A metabolism.  Also, it is unclear whether repeated administration of lipid nanoparticles will generate an immunologic response.  Nonetheless, the trial is an encouraging start on selective suppression of hepatic protein synthesis.

Propofol and sepsis

Motivation: Sepsis is one of the most common issues in the ICU, and for any number of reasons, patients with sepsis also could require intubation and sedation. Practically speaking, propofol and midazolam are both commonly used in the ICU for septic patients who require sedation, weighing their risks and benefits in the individual patient. There are some benefits of propofol over midazolam in critical care patients in general (1), so I wondered whether there were risks associated with using propofol in the setting of sepsis. This discussion excludes the issue of contaminated propofol preparations. I found one paper that raised clinical concerns about the association between propofol and infection/sepsis.

Study:  Haddad S, Tamim H, Memish ZA, Arabi Y. Association of preservative-free propofol use and outcome in critically ill patients. Am J Infect Control. 2011 Mar;39(2):141-7. 

Design: Nested cohort study of patients who were enrolled in a randomized control trial on another topic (comparing different regimens of insulin therapy). Patients who were on propofol were compared to those who were not on propofol, in terms of baseline characteristics and outcomes. ICU-acquired infections were defined as those found at least 48 hours after ICU admission and up to 48 hours after discharge, while ICU-acquired sepsis was defined as per standard criteria (2).

Results: There were differences in baseline characteristics between the two groups of patients; compared to no propofol, propofol group were more likely to be mechanically ventilated but less likely to have severe illness, chronic renal disease, immunosuppression or cardiovascular disease. In adjusted statistical analysis, propofol use was associated with increased risk of ICU-acquired infection (95 CI 1.17-3.05, p=0.009), and ICU-acquired severe sepsis and septic shock (95 CI 1.12-3.28,p=0.02). There were no statistically significant differences in ICU or hospital length of stay or ICU or hospital mortality (2). 

Discussion: This study had convincing evidence as above for a relationship between propofol use and infections, and discusses potential reasons why propofol may be associated with increased risk of infectious including impairment of host immune cell function, increased lipid levels affecting mitochondrial oxygen use, growth of bacteria in propofol emulsion, increased lipid caloric intake (2); however, it is important to emphasize that these are potential reasons based on studies or observations in other contexts that are not necessarily generalizable to the septic ICU patient. In terms of study limitations, it is important to note that the study showed a statistical association (not causation) between use of propofol and increased risk of ICU-acquired infection or sepsis in a post-hoc study on a trial designed for another indication (2). Noted that this study did not compare propofol versus midazolam, nor did I find a similarly designed study for midazolam.

I was surprised to see that there were several animal studies that showed benefit of propofol (versus no propofol) in animal models of sepsis. A few examples of these studies are here; most of the sample sizes are modest. In an animal study specifically designed to study the effect of propofol on a rat model of sepsis (cecal ligation and puncture), rats treated with early or late propofol had statistically significant higher survival rates than rats without propofol treatment (3). Propofol-treated rats also had statistically significant lower levels of ALT, AST, BUN, Cr and CK (3). In the same rat model of sepsis, rats treated with propofol had improved hypotension, lower plasma levels of TNF-alpha, IL-6, and more suppression of NF-kappaB activation (4). Another study reported that midazolam had an overall anti-inflammatory effect in a macrophage cell line through various mechanisms (5).

Looking for more studies of animal models of sepsis and sedation, I found a paper on the same rat model of sepsis comparing the effects of propofol and midazolam on neutrophil function (6). In both early and late time points of sepsis, propofol and midazolam decreased hydrogen peroxide production by neutrophils with propofol associated with statistically signficantly more decreased production than midazolam (6). These results suggest at least in the in vitro context that propofol might attenuate neutrophil function in this rat model of sepsis, favoring midazolam over propofol if results could be extrapolated to clinical context (6). Also, it was very interesting to note in this study that there was negligible effect of the lipid propofol carrier on neutrophil hydrogen peroxide production (6).

Conclusion: Overall, there lacks definitive evidence comparing propofol and midazolam in ICU patients with sepsis; however, one clinical study brought up concerning associations between propofol and ICU-related infection (2), while one animal model study showed that midazolam would be better than propofol in rat models of sepsis as it interfered less with neutrophil function (6). These concerns should be taken into account when weighing the risks and benefits of midazolam versus propofol for the septic ICU patient.

References:

1.     Carrasco G, Molina R, Costa J, Soler JM, Cabré L. Propofol vs midazolam in short-, medium-, and long-term sedation of critically ill patients. A cost-benefit analysis. Chest. 1993 Feb;103(2):557-64.

2.     Haddad S, Tamim H, Memish ZA, Arabi Y. Association of preservative-free propofol use and outcome in critically ill patients. Am J Infect Control. 2011 Mar;39(2):141-7.

3.     Bao HG, Li S. Effects of propofol on the outcomes of rats with sepsis. J Surg Res. 2011 Jun 1;168(1):e111-5. doi: 10.1016/j.jss.2010.12.034. Epub 2011 Jan 22.

4.     Song XM, Wang YL, Li JG, Wang CY, Zhou Q, Zhang ZZ, Liang H. Effects of propofol on pro-inflammatory cytokines and nuclear factor kappaB during polymicrobial sepsis in rats. Mol Biol Rep. 2009 Nov;36(8):2345-51. doi: 10.1007/s11033-009-9456-z. Epub 2009 Feb 4.

5.     Kim SN, Son SC, Lee SM, Kim CS, Yoo DG, Lee SK, Hur GM, Park JB, Jeon BH. Midazolam inhibits proinflammatory mediators in the lipopolysaccharide-activated macrophage. Anesthesiology. 2006 Jul;105(1):105-10.

6.     Inada T, Taniuchi S, Shingu K, Kobayashi Y, Fujisawa J, Nakao S. Propofol depressed neutrophil hydrogen peroxide production more than midazolam, whereas adhesion molecule expression was minimally affected by both anesthetics in rats with abdominal sepsis. Anesth Analg. 2001 Feb;92(2):437-41.

Abdominal pain and seizures

Motivation: A man walks in to the hospital with seizures also complains of periodic stomach pain.  I wonder could it also be a seizure?  After all, migraines in children are clearly linked to episodes of abdominal pain.  Could epileptic discharges create a similar phenomenon?

Turns out that this issue has been thought of by many - in fact, Trousseau had also thought about the issue in 1868.  Reviewed here is a compilation of the common clinical presentation.

Paper: Zinkin, NT, Peppercorn, MA. "Abdominal epilepsy" Best Practice & Research Clinical Gastroenterology (2005); 19(2): 263-274.

Methods: Review of 36 reported cases in english literature.  Patients were required to have paroxysmal EEG abnormalities with correlated clinical spells.

Results:
Clinical Characteristics: In meta-analysis of 36 published cases, most frequent feature was a sharp pain (86%) commonly in peri-umbilical and upper abdominal pain.  Nausea and vomiting was an infrequent feature in 28% followed by diarrhea (5%) and bloating (3%).  Besides abdominal pain, some disturbance in consciousness was observed in 64% ranging from generalized seizures to dizziness.  Duration of seizures was at most a few minutes.  EEG did not have a consistent localization for inter-ictal epileptiform activity.

Discussion: I think that abdominal epilepsy is worth considering as a diagnosis in someone with bouts of brief abdominal pain with some change in consciousness/"funny" feeling.  There are many people - often hard to treat people - with unexplained abdominal pain.  Some get unsatisfactory diagnosis of irritable bowel syndrome or somatoform disorders.  If there is no other explanation, an EEG may be in order.