1094
GASTROENTEROLOGY Vol. 102, No. 3
CORRESPONDENCE
Schreiber S, MacDermott RP, Raedler A, Pinnau R, Bertovich MJ, Nash GS. Increased activation of isolated intestinal lamina propria mononuclear cells in inflammatory bowel disease. Gastroenterology 1991;101:1020-1030. Elitsur Y, Freedland CP, Luk GD. Inhibition of DNA synthesis in human peripheral and lamina propria lymphocytes by 5-Aminosalicylic acid and hydrocortisone. Reg Immunol 1990;3:56-61. Hawthorne AB, Hawkey CJ. Immunosuppressive drugs in inflammatory bowel disease. A review of their mechanisms of efficacy and place in therapy. Drugs 1989;38:267-288.
Reply. I would like to thank Dr. Elitsur for his comments on our recent article describing the increased activation state of lymphocytes from patients with IBD. I would make a number of points in response to Dr. Elitsur’s letter. The system which we used was pokeweed mitogen activation of peripheral blood lymphocytes. In our previous manuscript, published in GASTROENTEROLOGY,l Figure 3, shows that 5-ASA at 0.5 mmol/L, 1 mmol/L, and 2 mmol/L concentrations showed no inhibition of pokeweed mitogen-induced mitogenesis. At the 3 mmol/L concentration, there was a slight 18% suppression of [3H]thymidine incorporation. In contrast, as shown in our recent manuscript published in GASTROENTEROLOGY,’ 1 mmol/L and 2 mmol/L concentration of 5-ASA caused 5O%-70% inhibition of activation marker expression (Interleukin 2 receptor and transferrin receptor). Therefore, at the 1 mmol/L and 2 mmol/L 5-ASA concentrations, using pokeweed mitogen-stimulated normal human peripheral blood mononuclear cells, we found that 5-ASA inhibited the activation marker expression but not mitogenesis. Furthermore, activation antigen marker expression on mononuclear cells represents an event distinctly different from cell proliferation and [3H]thymidine incorporation. In fact, in most systems, activation marker expression represents early events in response to antigen or mitogen stimulation that preceed in time subsequent cellular proliferation events. Dr. Elitsur’s publication represents a very different phenomenon in that he looked at Con A stimulation of lamina propria mononuclear cells. It is of interest to note in his paper that although intestinal lamina propria mononuclear cell mitogenesis was indeed markedly inhibited by 4 mmol/L 5ASA, the inhibition of peripheral blood lymphocyte Con A stimulation was a great deal less [approximately 32% with 4 mmol/L 5-ASA and 22% with 1 mmol/L 5-ASA). The fact that Dr. Elitsur used a different lectin for stimulation, employed higher doses of 5-ASA, and used lamina propria mononuclear cells explains some of the differences in conclusions between his studies and ours. In addition, it should be pointed out that the actual tissue concentration of the drug is particularly important. It is not known at this time what 5-ASA concentrations might be achieved within the inflamed intestine after ingestion of Azulfidine, Dipentum, Asacol, or Pentasa. Depending upon the amount of edema, extent, and depth of the inflammatory process, and changes in gastrointestinal motility or extent of diarrhea, the amount of 5-ASA present may not only be quite variable but also may be considerably lower than the 4 mmol/L dose utilized by Dr. Elitsur. Finally, with regard to the pathogenesis of IBD, I would point out that inhibition of cell activation with subsequent inhibition of functions such as cytokine production and/or adhesion molecule expression may have significant relevance to cellular events in the immunopathogenesis of IBD. In contrast, the inhibition of a nonspecific event such as lymphocyte proliferation may contribute little to prevention of intestinal damage. We thank Dr. Elitsur for his letter, for letting us know about his recent paper in Regional Immunology, and for discussing some of the effects on lymphocytes of 5-ASA.
RICHARD P. MACDERMOTT Department of Medicine University of Pennsylvania School of Medicine Philadelphia, Pennsylvania 1. MacDermott
RP, Schloemann SR, Bertovich MJ, Nash GS, Peters M, Stenson WF. Inhibition of antibody secretion by 5-aminosalicylic Acid. Gastroenterology 1989;96:442-448. 2. Schreiber S, MacDermott RP, Raedler A, Pinnau R, Bertovich MJ, Nash GS. Increased activation of isolated intestinal lamina propria mononuclear cells in inflammatory bowel disease. Gastroenterology 1991;101:1020-1030.
Statistical Methods for Predicting Variceal Hemorrhage Dear Sir: We read with great interest the paper by Kleber et al.’ The authors used Cox’s regression analysis to define the independent prognostic factors predicting variceal hemorrhage, and concluded that presence of gastric varices, alcoholic liver disease, presence of red color signs, and diameter of the largest varices were such factors. These results are relevant and important. However, the most important point is to define an index of risk of bleeding, and to assess the clinical usefulness of the index by measuring its sensitivity and specificity in predicting bleeding or death. Cox’s analysis is a powerful statistical tool, and its usefulness lies in the possibility of deriving prognostic indexes’ that may be used to assess in the individual patient the risk of variceal bleeding. Consequently, individual prognostic indexes may be used to define the best cut-off value for the prediction of variceal bleeding,3 and a different scoring system may be statistically compared.4 In the past few years, a prognostic index predicting variteal bleeding in cirrhotics without previous bleeding, the so-called NIEC score, was developed by an Italian multicenter group,’ and is currently used by many gastroenterological units as a clinical tool to define bleeding risk. According to a ROC curve analysis,4 it was shown to be superior to a classification based only on variceal size.6 We think that the authors should go on with their analysis and calculate prognostic indexes, assess by ROC curve analysis the efficiency of their index, and compare it with the NIEC score or with the conventional classification of variceal size. If an improvement in efficiency results, this observation would support the conclusion that the assessment of the parameters included in the prognostic index developed by the authors is clinically useful in the prediction of hemorrhage in patients with cirrhosis without previous bleeding. CARLO MERKEL ANGELO GATTA Department of Clinical Medicine University of Padua Padua, Italy Kleber G, Sauerbruch T, Ansari H, Paumgartner G. Prediction of variceal hemorrhage in cirrhosis. A prospective follow-up study. Gastroenterology 1991;100:1332-1337. Christensen E. Multivariate survival analysis using Cox’s regression models. Hepatology 1987;7:1346-1358. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curve derived from the same cases. Radiology 1983;148:839-843. Hanley JA, McNeil BJ. The meaning and use of the area under a
COKRESPONDENCE
March 1992
1095
ing risk score including the parameters that had been proven to be independent prognostic indicators of first variceal bleeding.’ This score takes into account the relative importance of the different variables. When applied to Figure 1,the score can be used to estimate the first bleeding risk within a given time for individual patients. This score should be evaluated prospectively in a different cohort of patients with respect to sensitivity and specificity including the use of ROC curves as suggested. We agree that comparison with different scores, e.g., the NIEC score,’ is important. Thus, for validation and comparison we recommend the use of both scores in future studies. Based on the formula of the Cox’s model and the regression coefficients of the different variables our prognostic score was calculated as
receiver operating characteristic (HOC) curve. Radiology 1982;143:29-36. North Italian Endoscopic Club for the study and treatment of esophageal varices. Prediction of the first variceal hemorrhage in patients with cirrhosis of the liver and esophageal varices. N Engl J Med 1988;319:983-989. Merkel C, Gatta A. Can we predict the first variceal bleeding in the individual patient with cirrhosis and esophageal varices? (letter). J Hepatol 1991;12:402. Reply.
We thank Drs. Merkel and Gatta for their valuable comments. As suggested, we calculated from the Cox’s model a bleed-
Prognostic score = exp(l.7019*GV
+ l.O483*KCS + 0.8158*ALD + 0.25Yl:SIZE).
I-
Score
0.0
I
O
’
I
m
I
I
’
’
I
20 30 40 50 10 Months after inclusion
n
For calculation of the individual score for a given patient gastric varices in the fundus (GV), red color sign (RCS), alcoholic liver disease (ALD) and SIZE (esophageal variceal diameter r5 mm) must be substituted with 0 and 1, where 0 denotes absence of the respective sign and 1 denotes presence of the respective sign. The higher the value of the index the higher is the bleeding risk. To facilitate calculation, we include a pocket chart (Table I), showing the calculated index for each possible combination of variables. Using the calculated index for a given patient, the bleeding probability within a given time can be estimated from the figure or the table. Provided that future prospective studies can validate our index, the figure and pocket chart may useful advice for assessing the bleeding risk of a particular patient.
>5
GEKHAKII KLEBEK HASAN ANSARI TILMAN SAUERBKUCH
1
60
Deportment of Medicine II Klinikum Grosshodern, University of Munich Munich, Gernmny
Figure 1. Probability to have no variceal bleeding as a function of time and bleeding risk score (survival function, Cox’s model]. Score was obtained as described from analysis of 106 patients with varices and no previous variceal bleeding.’ The three curves show the probability of having no variceal bleeding in groups of patients with different risk scores (score ~2: n = 31, score 2-5: n = 52, score > 5: n = 26). The curves were significantly different (x2 = 24.1, P < 0.0001).
Table 1. Pocket Chart for Easy Calculation Alcoholic liver disease Gastric varices (fundus) Variceal Size 2 5 mm Red color sign Index Bleeding risk 6 months 12 months 24 months
-
1.
Kleber G, Sauerbruch T, Ansari H, Paumgartner G. Prediction of variceal hemorrhage in cirrhosis. A prospective follow-up study. Gastroenterology 1991;100:1332-1337. 2. North Italian Endoscopic Club for the study and treatment of esophageal varices. Prediction of the first variceal hemorrhage in patients with cirrhosis of the liver and esophageal varices. N Engl J Med 1988;319:983-989.
of the Bleeding Risk Score and for Estimation
0 0 0 0
0 0 1 0
5
0.43 0.62 0.66
,
GASTROENTEROLOGY Vol. 102, No. 3
CORRESPONDENCE
Schreiber S, MacDermott RP, Raedler A, Pinnau R, Bertovich MJ, Nash GS. Increased activation of isolated intestinal lamina propria mononuclear cells in inflammatory bowel disease. Gastroenterology 1991;101:1020-1030. Elitsur Y, Freedland CP, Luk GD. Inhibition of DNA synthesis in human peripheral and lamina propria lymphocytes by 5-Aminosalicylic acid and hydrocortisone. Reg Immunol 1990;3:56-61. Hawthorne AB, Hawkey CJ. Immunosuppressive drugs in inflammatory bowel disease. A review of their mechanisms of efficacy and place in therapy. Drugs 1989;38:267-288.
Reply. I would like to thank Dr. Elitsur for his comments on our recent article describing the increased activation state of lymphocytes from patients with IBD. I would make a number of points in response to Dr. Elitsur’s letter. The system which we used was pokeweed mitogen activation of peripheral blood lymphocytes. In our previous manuscript, published in GASTROENTEROLOGY,l Figure 3, shows that 5-ASA at 0.5 mmol/L, 1 mmol/L, and 2 mmol/L concentrations showed no inhibition of pokeweed mitogen-induced mitogenesis. At the 3 mmol/L concentration, there was a slight 18% suppression of [3H]thymidine incorporation. In contrast, as shown in our recent manuscript published in GASTROENTEROLOGY,’ 1 mmol/L and 2 mmol/L concentration of 5-ASA caused 5O%-70% inhibition of activation marker expression (Interleukin 2 receptor and transferrin receptor). Therefore, at the 1 mmol/L and 2 mmol/L 5-ASA concentrations, using pokeweed mitogen-stimulated normal human peripheral blood mononuclear cells, we found that 5-ASA inhibited the activation marker expression but not mitogenesis. Furthermore, activation antigen marker expression on mononuclear cells represents an event distinctly different from cell proliferation and [3H]thymidine incorporation. In fact, in most systems, activation marker expression represents early events in response to antigen or mitogen stimulation that preceed in time subsequent cellular proliferation events. Dr. Elitsur’s publication represents a very different phenomenon in that he looked at Con A stimulation of lamina propria mononuclear cells. It is of interest to note in his paper that although intestinal lamina propria mononuclear cell mitogenesis was indeed markedly inhibited by 4 mmol/L 5ASA, the inhibition of peripheral blood lymphocyte Con A stimulation was a great deal less [approximately 32% with 4 mmol/L 5-ASA and 22% with 1 mmol/L 5-ASA). The fact that Dr. Elitsur used a different lectin for stimulation, employed higher doses of 5-ASA, and used lamina propria mononuclear cells explains some of the differences in conclusions between his studies and ours. In addition, it should be pointed out that the actual tissue concentration of the drug is particularly important. It is not known at this time what 5-ASA concentrations might be achieved within the inflamed intestine after ingestion of Azulfidine, Dipentum, Asacol, or Pentasa. Depending upon the amount of edema, extent, and depth of the inflammatory process, and changes in gastrointestinal motility or extent of diarrhea, the amount of 5-ASA present may not only be quite variable but also may be considerably lower than the 4 mmol/L dose utilized by Dr. Elitsur. Finally, with regard to the pathogenesis of IBD, I would point out that inhibition of cell activation with subsequent inhibition of functions such as cytokine production and/or adhesion molecule expression may have significant relevance to cellular events in the immunopathogenesis of IBD. In contrast, the inhibition of a nonspecific event such as lymphocyte proliferation may contribute little to prevention of intestinal damage. We thank Dr. Elitsur for his letter, for letting us know about his recent paper in Regional Immunology, and for discussing some of the effects on lymphocytes of 5-ASA.
RICHARD P. MACDERMOTT Department of Medicine University of Pennsylvania School of Medicine Philadelphia, Pennsylvania 1. MacDermott
RP, Schloemann SR, Bertovich MJ, Nash GS, Peters M, Stenson WF. Inhibition of antibody secretion by 5-aminosalicylic Acid. Gastroenterology 1989;96:442-448. 2. Schreiber S, MacDermott RP, Raedler A, Pinnau R, Bertovich MJ, Nash GS. Increased activation of isolated intestinal lamina propria mononuclear cells in inflammatory bowel disease. Gastroenterology 1991;101:1020-1030.
Statistical Methods for Predicting Variceal Hemorrhage Dear Sir: We read with great interest the paper by Kleber et al.’ The authors used Cox’s regression analysis to define the independent prognostic factors predicting variceal hemorrhage, and concluded that presence of gastric varices, alcoholic liver disease, presence of red color signs, and diameter of the largest varices were such factors. These results are relevant and important. However, the most important point is to define an index of risk of bleeding, and to assess the clinical usefulness of the index by measuring its sensitivity and specificity in predicting bleeding or death. Cox’s analysis is a powerful statistical tool, and its usefulness lies in the possibility of deriving prognostic indexes’ that may be used to assess in the individual patient the risk of variceal bleeding. Consequently, individual prognostic indexes may be used to define the best cut-off value for the prediction of variceal bleeding,3 and a different scoring system may be statistically compared.4 In the past few years, a prognostic index predicting variteal bleeding in cirrhotics without previous bleeding, the so-called NIEC score, was developed by an Italian multicenter group,’ and is currently used by many gastroenterological units as a clinical tool to define bleeding risk. According to a ROC curve analysis,4 it was shown to be superior to a classification based only on variceal size.6 We think that the authors should go on with their analysis and calculate prognostic indexes, assess by ROC curve analysis the efficiency of their index, and compare it with the NIEC score or with the conventional classification of variceal size. If an improvement in efficiency results, this observation would support the conclusion that the assessment of the parameters included in the prognostic index developed by the authors is clinically useful in the prediction of hemorrhage in patients with cirrhosis without previous bleeding. CARLO MERKEL ANGELO GATTA Department of Clinical Medicine University of Padua Padua, Italy Kleber G, Sauerbruch T, Ansari H, Paumgartner G. Prediction of variceal hemorrhage in cirrhosis. A prospective follow-up study. Gastroenterology 1991;100:1332-1337. Christensen E. Multivariate survival analysis using Cox’s regression models. Hepatology 1987;7:1346-1358. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curve derived from the same cases. Radiology 1983;148:839-843. Hanley JA, McNeil BJ. The meaning and use of the area under a
COKRESPONDENCE
March 1992
1095
ing risk score including the parameters that had been proven to be independent prognostic indicators of first variceal bleeding.’ This score takes into account the relative importance of the different variables. When applied to Figure 1,the score can be used to estimate the first bleeding risk within a given time for individual patients. This score should be evaluated prospectively in a different cohort of patients with respect to sensitivity and specificity including the use of ROC curves as suggested. We agree that comparison with different scores, e.g., the NIEC score,’ is important. Thus, for validation and comparison we recommend the use of both scores in future studies. Based on the formula of the Cox’s model and the regression coefficients of the different variables our prognostic score was calculated as
receiver operating characteristic (HOC) curve. Radiology 1982;143:29-36. North Italian Endoscopic Club for the study and treatment of esophageal varices. Prediction of the first variceal hemorrhage in patients with cirrhosis of the liver and esophageal varices. N Engl J Med 1988;319:983-989. Merkel C, Gatta A. Can we predict the first variceal bleeding in the individual patient with cirrhosis and esophageal varices? (letter). J Hepatol 1991;12:402. Reply.
We thank Drs. Merkel and Gatta for their valuable comments. As suggested, we calculated from the Cox’s model a bleed-
Prognostic score = exp(l.7019*GV
+ l.O483*KCS + 0.8158*ALD + 0.25Yl:SIZE).
I-
Score
0.0
I
O
’
I
m
I
I
’
’
I
20 30 40 50 10 Months after inclusion
n
For calculation of the individual score for a given patient gastric varices in the fundus (GV), red color sign (RCS), alcoholic liver disease (ALD) and SIZE (esophageal variceal diameter r5 mm) must be substituted with 0 and 1, where 0 denotes absence of the respective sign and 1 denotes presence of the respective sign. The higher the value of the index the higher is the bleeding risk. To facilitate calculation, we include a pocket chart (Table I), showing the calculated index for each possible combination of variables. Using the calculated index for a given patient, the bleeding probability within a given time can be estimated from the figure or the table. Provided that future prospective studies can validate our index, the figure and pocket chart may useful advice for assessing the bleeding risk of a particular patient.
>5
GEKHAKII KLEBEK HASAN ANSARI TILMAN SAUERBKUCH
1
60
Deportment of Medicine II Klinikum Grosshodern, University of Munich Munich, Gernmny
Figure 1. Probability to have no variceal bleeding as a function of time and bleeding risk score (survival function, Cox’s model]. Score was obtained as described from analysis of 106 patients with varices and no previous variceal bleeding.’ The three curves show the probability of having no variceal bleeding in groups of patients with different risk scores (score ~2: n = 31, score 2-5: n = 52, score > 5: n = 26). The curves were significantly different (x2 = 24.1, P < 0.0001).
Table 1. Pocket Chart for Easy Calculation Alcoholic liver disease Gastric varices (fundus) Variceal Size 2 5 mm Red color sign Index Bleeding risk 6 months 12 months 24 months
-
1.
Kleber G, Sauerbruch T, Ansari H, Paumgartner G. Prediction of variceal hemorrhage in cirrhosis. A prospective follow-up study. Gastroenterology 1991;100:1332-1337. 2. North Italian Endoscopic Club for the study and treatment of esophageal varices. Prediction of the first variceal hemorrhage in patients with cirrhosis of the liver and esophageal varices. N Engl J Med 1988;319:983-989.
of the Bleeding Risk Score and for Estimation
0 0 0 0
0 0 1 0
5
0.43 0.62 0.66
,
