This Month in Gastroenterology

Article Outline

• Risks of Undiagnosed Celiac Disease
• Inflammatory Bowel Disease and Risk for Recurrent Thromboembolism
• A Discrete Natural Killer Cell Population Activated for γ-Interferon Production by Macrophages in Crohn's Disease
• A Cause for Pruritis in Chronic Liver Disease
• Copyright

Back to Article Outline

Risks of Undiagnosed Celiac Disease 

Celiac disease is a well-recognized cause of malabsorptive diarrhea, which responds to a strict, gluten-free diet. Untreated symptomatic celiac disease is associated with substantial morbidity and increased mortality. The availability of noninvasive, serologic testing has revealed a higher prevalence of asymptomatic or minimally symptomatic celiac disease than previously appreciated. The impact of undiagnosed, and therefore untreated, celiac disease on morbidity or mortality is currently uncertain.

In this issue of Gastroenterology, Godfrey et al report the results of a study investigating the morbidity and mortality of undiagnosed celiac disease in a population-based sample of subjects age ≥50. As part of a prior study, serum samples of 24 727 Olmsted County residents age ≥50 were obtained between 1995 and 2001 and stored. Samples from 16 886 subjects who had given consent and had no known celiac disease were tested for tissue transglutaminase antibodies (tTGA) and endomysial antibodies (EMA). Undiagnosed CD was defined by the presence of a tTGA level >2.0 U/mL with a positive EMA test. Tests were considered indeterminate if the tTGA level was >4.0 U/mL and the EMA test was negative. After completion of serologic testing, a nested, matched, case-control design was proposed to compare undiagnosed celiac disease subjects with controls based on 2:1 matching of age and gender. Complete medical records of the subjects were reviewed for comorbidity and mortality by individuals unaware of serum status.

A total of 129 subjects (51% women; median age, 63 years) showed a seropositive result, resulting in a population prevalence of undiagnosed celiac disease of 0.8% (95% confidence interval, 0.6%–0.9%). A total of 254 matched controls were selected for comparison. Compared with controls, celiac disease was associated with lumbar spine T-scores, an increased risk of osteoporosis (−1.7 vs −0.9; odds ratio, 0.64; 95% confidence interval, 0.48–0.85) and a greater prevalence of hypothyroidism. Undiagnosed celiac disease patients had lower weight and body mass index values (median value, 26.4 vs 27.4), reduced plasma cholesterol (median value, 200.0 vs 213.0) and ferritin (25.0 vs 78.5) levels, and tended to have lower rates of glucose intolerance.

There was also no significant association of undiagnosed celiac disease with irritable bowel syndrome, diarrhea, other gastrointestinal symptoms, or unexplained weight loss. No significant increase in cancer risk was found in undiagnosed celiac disease (Figure 1). The rates of all-cause mortality (hazard ratio, 0.80; 95% confidence interval, 0.45–1.41) or cancer mortality did not differ between undiagnosed celiac disease and controls. Twenty patients (15 women) with undiagnosed celiac disease, and none of the controls, were later diagnosed with celiac disease, mainly based on the presence of iron deficiency or dermatitis herpetiformis.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 1. 

  Summary of outcomes of undiagnosed celiac disease cases compared with serology-negative controls. Odds ratio (95% CI) from conditional logistic regression.

This population-based study shows that undiagnosed celiac disease is associated with an increased rate of osteoporosis and lower bone density scores, but not with increased mortality or other major comorbidities and symptoms commonly linked to diagnosed CD. At least for the period of the follow-time that was assessed in this study, detection of undiagnosed celiac disease is unlikely to generate a major health benefit.

See page 763.

Back to Article Outline

Inflammatory Bowel Disease and Risk for Recurrent Thromboembolism 

It is well-known that patients with inflammatory bowel disease (IBD) have an increased risk of venous thromboembolism and major complications such as pulmonary embolism. Anticoagulant treatment is started at the time of thromboembolic events, and the required therapy duration depends on the risk of recurrent thromboembolism. However, the rate and risk factors associate with recurrence of venous thromboembolism in IBD have not been established.

In this issue of Gastroenterology, Novacek et al report the results of a prospective study that compared the risk of recurrent thromboembolism in patients with IBD with the risk in patients without IBD. For the IBD cohort, IBD patients from 14 referral centers in Austria were retrospectively analyzed for a history of venous thromboembolism of the lower or upper limbs, or pulmonary embolism. Only thromboembolism confirmed by objective imaging techniques were taken into account. Thromboembolisms occurring secondary to trauma, surgery, or pregnancy were classified as “provoked venous thromboembolisms”; all others were classified as “unprovoked venous thromboembolisms.” A detailed IBD history was obtained from all cases. Patients entered the study on the last day of anticoagulant therapy. The primary end point was recurrence of venous thromboembolism after discontinuation of anticoagulant therapy. The control cohort consisted of a group of patients with a first venous thromboembolism, attending 4 thrombosis centers in Vienna, who had no history of IBD.

A total of 142 patients with a first venous thromboembolism after a diagnosis of IBD were identified. In 116 of these (mean age, 42 years; 55% female), anticoagulation was stopped after on average 6 months treatment, with a mean observation time of 42 months. Thromboembolism was provoked by surgery or trauma in 26%. Recurrent venous thromboembolism, the vast majority unprovoked, occurred in 35 patients (30%). In a multivariate model, independent risk factors for recurrent venous thromboembolism were male gender and age at the first thromboembolic event.

The 86 IBD patients with unprovoked venous thromboembolism were compared to a control group without IBD (n = 1255; mean age, 48 years; 51% female). The inflammatory bowel disease group was significantly younger, had a lower body mass index, a shorter duration of anticoagulation, more often a thrombosis of the proximal leg veins, and higher factor VIII levels. The probability of recurrence after 5 years was 33.4% among IBD patients with unprovoked first venous thromboembolism and 21.7% among patients without IBD (P = .01; Figure 2). After adjustment for potential confounders, IBD was an independent risk factor of recurrence (hazard ratio, 2.5; 95% confidence interval, 1.4–4.2; P = .001).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 2. 

  Kaplan–Meier estimates of recurrent venous thromboembolism after a first unprovoked venous thromboembolism in patients with (IBD) and without IBD (controls).

This study confirms that IBD patients are at high risk of recurrent thrombosis after a first venous thromboembolism. Study limitations are the lack of systematic thrombophilia screening in all patients, the recruitment from referral centers, and the retrospective nature of the analysis. These findings might argue for longer term anticoagulant treatment after venous thromboembolism in IBD patients, although further studies will be needed to evaluate the risk–benefit ratio of prolonged therapy in IBD.

See page 779.

Back to Article Outline

A Discrete Natural Killer Cell Population Activated for γ-Interferon Production by Macrophages in Crohn's Disease 

Crohn's disease may represent an exaggerated or improper immune response from the innate immune system. Specific innate immune cells, such as natural killer (NK) cells that are present in the mucosa and are distinct from conventional NK cells by secreting interleukin (IL)-22, might participate in pathologic or protective processes for chronic inflammation in the gut.

In the study by Takayama et al, isolated CD3⁻CD56⁺ NKp44⁺NKp46⁻ and NKp44⁻NKp46⁺ NK cells from lamina propria mononuclear cells were utilized from normal and Crohn's disease intestinal specimens to examine their interaction with isolated CD14⁺ lamina propria macrophages and to study their potential role in the pathogenesis of Crohn's disease. Unlike peripheral blood NK cells, which are IL-15–dependent conventional NK cells, lamina propria NK cells showed a novel dimorphic population: 1 expressing NKp44⁺ (also expressing CD127 but not CD122, as well as RORC and IL-22) and 1 expressing NKp46⁺ (also expressing CD122 but not CD127, and the IL-23 receptor). The proportion of NKp44⁺ NK cells were significantly decreased in Crohn's disease, whereas the proportion of NKp46⁺ was significantly elevated compared with samples from normal and ulcerative colitic intestines. The NKp46⁺ NK cells, possessing the IL-23 receptor, responds via direct cell-to-cell contact with CD14⁺ lamina propria macrophages that produce IL-23, also increased in Crohn's disease intestines, and stimulates NKp46⁺ NK cells to produce copious amounts of interferon (IFN)-γ (Figure 3). Depletion of CD14⁺ lamina propria macrophages abolished the commensal bacterial effect on IFN-γ production by NKp46⁺ NK cells, as did separation of the interaction between NKp46⁺ NK cells and CD14⁺ lamina propria macrophages, and recombinant IL-23 alone did not replicate the response.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 3. 

  Direct cell-to-cell interaction between lamina propria (LP) NK cells and LP CD14⁺ macrophages (Co = co-culture) that produce IL-23 is required to produce IFN-γ protein. With commensal bacterial stimulation (E faecalis), co-culture causes a marked increase in IFN-γ production that was blocked with anti–IL-23 antibodies (α23). iso, non-IL23 antibody; tw, tap water control.

The study indicates that an imbalance exists between unique CD3⁻CD56⁺ NKp44⁺NKp46⁻ and NKp44⁻NKp46⁺ NK cells in the lamina propria of Crohn's patients in favor of NKp46⁺ NK cells. NKp46⁺ NK cells secrete IFN-γ upon direct interaction with CD14⁺ macrophages and its macrophage-produced IL-23 to stimulate the NKp46⁺ cells. This unique population of NK cells is involved in the pathogenesis of Crohn's disease.

See page 882.

Back to Article Outline

A Cause for Pruritis in Chronic Liver Disease 

The chemical mediator for pruritis, the intense itching associated best with cholestatic liver disease, is not known. Several potential candidates, such as bile salts, histamine, and endogenous opiod peptides, have not correlated well with the severity of pruritis.

In the study by Kremer et al, sera from pruritic patients with hepatic cholestasis and a mouse model for pruritis were utilized to determine the etiology of the pruritic molecule. By using intracellular calcium changes as an indicator for neuronal activation, sera from pruritic patients were screened for calcium activation when added to SH-SY5Y neuroblastoma cells. Utilizing several sera pretreatment strategies to help determine the composition of the pruritic agent, it was discovered that it was not a peptide (not sensitive to 90% ethanol or proteinase K), could pass a 100-kD but not a 10-kD filter, and could be totally recovered through a 10-kD filter with the addition of cholate above its critical micellular concentration, suggesting that the molecule was a small hydrophobic substance, and showed increased hydrophobicity on acidification, suggesting a small phospholipid that was activating increased intracellular calcium. Lysophosphatidic acid (LPA) emerged as the candidate because it (a) increases intracellular calcium in neuronal cells that could be blocked with a specific LPA receptor blocker, and (b) was markedly elevated in the cholestatic sera from patients. Because LPA is formed in blood through autotaxin (ATX)-induced cleavage of lysophosphatidylcholine, the amount of LPA in sera primarily depends on ATX activity. Indeed, ATX activity was higher in cholestatic patients, and specifically was highest in those patients suffering from pruritis. A highly significant correlation was observed between ATX enzyme activity and the intensity of pruritis on a visual analog scale. Nasobiliary drainage, which reduced ATX activity by an unknown mechanism because ATX is not found in bile, matched the improvement of pruritis in patients, and both activity as well as pruritis returned after nasobiliary drainage ceased (Figure 4). This correlation was not seen for histamine, tryptase, substance P, bile salts, and μ-opiod activity. Using a mouse model in which scratch activity can be measured, intradermal LPA induced a dose-dependent pruritis.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 4. 

  Only ATX activity, but not histamine, nor tryptase, substance P, serum bile salts or μ-opiod activity in serum correlated with itch intensity of patients with cholestatic itch. ATX activity showed a significant linear correlation with the itch intensity represented as a visual analogue scale ranging from 0 (no pruritis) to 10 (most severe form of pruritis; Spearman's correlation coefficient; r = 0.7764; P < .0001).

This study indicates that ATX activity and its product, LPA, is the likely mediator for pruritis in cholestatic liver disease, and probably triggers pruritoceptive nerve fibers by activating intracellular calcium release. This discovery could lead to clinical inhibitors of ATX and LPA to decrease morbidity from pruritis in liver-related and other conditions.

See page 1008.