- Email: [email protected]
A prospective study of the ability of three endoscopic classifications to predict hemorrhage from esophageal varices
0016-5107/92/3804-0425$03.00 GASTROINTESTINAL ENDOSCOPY Copyright © 1992 by the American Society for Gastrointestinal Endoscopy
A prospective study of the ability of three endoscopic classifications to predict hemorrhage from esophageal varices G. P. Rigo, MD, A. Merighi, MD N. J. Chahin, MD, M. Mastronardi, MD P. L. Codeluppi, MD, A. Ferrari, MD C. Armocida, MD, G. Zanasi, MD A. Cristani, MD, G. Cioni, MD F. Manenti Modena, Italy
Hemorrhage from esophageal varices in cirrhotics is a frequent event with high mortality in spite of therapy. Preventive sclerotherapy seems to be beneficial only if the patient's bleeding risk is higher than 40 to 50% a year. A series of 320 patients with esophageal varices without previous bleeding was studied prospectively; the varices were classified according to three widely used endoscopic classifications. During follow-up (6 to 36 months, average 14 months), hemorrhage occurred in 49 patients (15.3%) of whom 30 (61.2%) bled from varices (8.2 and 11.0% at 12 and 24 months, respectively). At the same time intervals, mortality of the entire population studied was 18.0 and 23.8%, respectively, of which one third was directly due to hemorrhage. With all three classifications, the higher the degree of bleeding risk, the greater the actual percentage of hemorrhages recorded; however, it never reached 40% a year. In predicting the bleeding event, Dagradi's classification proved more sensitive than JRSPH or NIEC, but the latter classifications were more specific and assessed a higher predictive value for a positive test. Endoscopic observation probably needs integration with other methods if a reliable bleeding prediction is to be made. (Gastrointest Endosc 1992;38:425-429)
Patients suffering from portal hypertension frequently hemorrhage from esophageal varices. This is a dramatic event accompanied by a high incidence of mortality.1 Analysis of controlled trials on different types of prophylactic treatment (drugs, endoscopy, surgery) against the first episode of bleeding from esophageal varices has shown that it is necessary to select patients with a high risk of bleeding in order to get the best results. 2 Many authors have observed that patients with large varices are more at risk than those with small varices. 3- s Beppu et al} in a retrospective
Received June 20, 1991. For revision August 15, 1991. Accepted December 26, 1991. From Gastroenterology and Digestive Endoscopy and 3rd Department of Clinical Medicine, University of Modena, Modena, Italy. Reprint requests: Gianpiero Rigo, MD, Cattedia di Gastroenterologia ed EndoscopiaDigestiva, Policlinico, viadelPozzo 71, 41100 Modena, Italy. VOLUME 38, NO.4, 1992
study, applied the criteria of the Japanese Research Society for Portal Hypertension lO (JRSPH) to patients with esophageal varices and were able to calculate a score which identified high-risk patients (80 to 100%). Paquet,!1 in a controlled trial of prophylactic sclerotherapy, selected patients exhibiting large varices with black points or large varices associated with reduced prothrombin activity (30% of normal values). In 2 years of follow-up, 66% of the controls suffered from variceal bleeding. In 1988, The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices (NIEC)12 proposed a prognostic index for hemorrhage from esophageal varices taking into account the size of the varices, the presence of red wale markings, and Child's classification. Using this index, the authors were able to define six risk classes, in the last of which bleeding occurred in 68.9% of the patients in 1 year of follow425
was classified as non-variceal but due to associated lesions if there was actual bleeding from the lesion (erosions and ulcers, benign or malignant lesions of the esophagus, stomach and duodenum and acute hemorrhagic gastritis) or stigmata from the Forrest classification. Otherwise, the bleeding was classified as being of unidentified origin. 16 Mortality was considered to be related to bleeding only when it occurred within 6 weeks from the hemorrhage and was linked to it by a continuous sequence of events as evident from clinical records of the hospital where the patients had been admitted. Bleeding episodes and mortality were evaluated by means of the life table methodY
up. A simplified index was introduced more recently 13 with substantially identical results. The aim of our prospective study was 3-fold: to calculate the incidence of upper digestive tract hemorrhage due to esophageal varices and from other bleeding sites, to determine total mortality and that related to bleeding; and to evaluate the possibility of identifying high-risk patients using three classifications of esophageal varices in a population of patients with esophageal varices but without a history of previous bleeding. PATIENTS AND METHODS
All patients with esophageal varices diagnosed endoscopically in the Gastrointestinal Unit of the University of Modena between October 1, 1984 and February 28, 1986 were considered for the study. The criteria for exclusion were: (1) the existence of previous bleeding episodes from esophageal varices or of unidentified origin; (2) previous treatment (pharmacological, surgical, endoscopic) capable of modifying portal pressure or the risk of hemorrhage; and (3) associated diseases (such as cancer) with reduced life expectancy. All of the patients' data were computerized: they included all parameters (clinical, biochemical, and endoscopic) necessary for the calculation of Child's classification (modified by Pugh)14 and of the esophageal varices grading according to the classifications of Dagradi,5 JRSPH, and NIEC. Dagradi5 suggested five degrees depending on the size of the varices, the fifth degree being characterized by large varices with cherry red spotS. 5,6 The JRSPH classification proposed six classes of risk based not only on the evaluation of the four fundamental parameters (color, form, and extension of the varices and red signs) but also on the presence or absence of esophagitis. 10 The NIEC index also described six classes of risk but based only on three parameters: Child's class, size of varices, and the presence of red wale markings. 12 In order to ensure maximal homogeneity in the description of the esophageal varices, all endoscopists participating underwent a preliminary training program set up as part of the Italian Liver Cirrhosis Project. 15 All relevant clinical events were recorded 6, 12, and 24 months after initial classification. When direct examination was not possible, the patients or their physicians were contacted by phone at the intervals mentioned above. When a hemorrhage was reported during follow-up, the source of bleeding, as ascertained by emergency endoscopy, was classified as variceal when the following were observed: bleeding from the varices, white nipples on the varices, nonbleeding fifth-degree varices, and no other potential or apparent sources of bleeding in the upper digestive tract. It
RESULTS
From October 1, 1984 to February 28, 1986, 477 patients with esophageal varices were examined by the endoscopists of the unit. A total of 136 patients were excluded from the study: 52 patients because of previous bleeding episodes from varices or other unidentified lesions; 64 patients owing to previous treatment capable of modifying portal pressure and bleeding risk; and 20 patients because of liver hepatoma or other severe associated disorders. In 21 patients the data were insufficient for complete evaluation. The remaining 320 patients were considered to be suitable for purpose of this study. Table 1 gives the demographic data and diagnosis of the patients. Follow-up continued for 6 to 36 months, the average was 14 months. Hemorrhage: incidence and source of bleeding
During follow-up, 49 patients (15.3%) bled (Table 2). The bleeding was due to varices in 30 patients (61.2%) and to associated lesions in 10 patients (20.4%); in 9 patients (18.4%) the source of bleeding was unidentified. Of the bleeding episodes due to esophageal varices (Fig. 1), the incidence rose from 8.2% at 12 months to 11.0% at 24 months. Mortality: incidence and causes
During follow-up, 63 patients died; 18.0% at 12 months and 23.8% at 24 months (Table 2). Hemorrhage accounted for approximately one third of total deaths: bleeding was due to varices in nine patients, to associated lesions in two patients, and was of unidentified origin in eight patients (these patients died
Table 1. Data of patients at entry Diagnosis No. of patients
320 426
Sex
M/F
Age (yr)
231/89 54.8 ± 11
Child classification
Liver Secondary biliary Portal cirrhosis cirrhosis thrombosis
318
1
1
A
B
C
108 125 64
Not determined
23
GASTROINTESTINAL ENDOSCOPY
Table 2. Cumulative rate of bleeding and mortality
Bleeding Overall Variceal Associated lesions Unidentified origin Mortality Overall Due to bleeding
Total no. of patients
Rate (%) 12 months
24 months
49 30 10 9
12.5 8.2 2.2 2.1
20.9 11.0 4.5 5.4
63 19
18.0 5.4
23.8 6.6
Table 3. Cumulative percentage (Kaplan-Meier method) of variceal bleeding among 320 patients with portal hypertension and esophageal varices classified according to Dagradi 5 • 6 at entry Varices degree VI V2 V3 V4 V5 Total
No.ofpatients who bled/total
12 months
24 months
0/5 5/122 3/82 4/26 18/85
3.6 2.7 13.0 19.8
3.6 2.7 19.0 24.0
30/320
8.2
11.0
Rate of bleeding (%)
100
j ~
!
t . ~
Table 4. Cumulative percentage (Kaplan-Meier method) of patients bleeding among 320 patients with portal hypertension and esophageal varices classified according to JRSPH'o at entry
80
70
Risk class
JRSPH score
No. of patients Rate of bleeding (%) who bled/total 12 months 24 months
60
50
oJ----r----,---.--
-'~7--30"-
12 -'Tis--18"-'i,--2'" r-',
1 2 3 4 5 6
months
Figure 1. Cumulative percentage of patients free of variceal bleeding during the study (life table).
in other hospitals where no emergency endoscopic examination was performed).
>1.14 0.38 to 1.14 0.38 to 0.0 0.0 to -0.38 -0.38 to -1.14 <-1.14 Total
3/131 7/88 3/15 5/34 6/28 6/24
2.5 8.1 6.9 13.4 16.4 24.0
2.5 8.1 6.9 13.4 21.2 31.6
30/320
8.2
11.0
Table 5. Cumulative percentage (Kaplan-Meier method) of patients bleeding among 297 patients with portal hypertension and esophageal varices classified according to NIEC'2 at entry
Bleeding risk from esophageal varices
Risk
NIEC in-
No. of patients
The cumulative incidence of hemorrhagic episodes at 12 and 24 months varied, depending on the classification of risk adopted: according to the classification of Dagradi,5 it was markedly higher in patients with fourth- and fifth-degree varices than in those with second- and third-degree varices (Table 3). According to JRSPH, it increased progressively as the class of risk rose, reaching a maximum of 24.0 and 31.6%, respectively, in patients at maximum risk (Table 4). According to the NIEC index, it was significantly lower in patients with a risk factor <35.0 (classes I, II, III, and IV) than it was in those with a risk factor >35.0 (classes V and VI). The high rate of bleeding at 12 months (37.9%) in patients of the sixth class, with no further increase at 24 months, is particularly worth noting (Table 5).
class
dex
who bled/total
12 months
24 months
<20 20.0-25.0 25.1-30.0 30.1-35.0 35.1-40.0 >40
2/50 5/94 6/48 3/38 7/48 5/19
2.1 6.1 4.3 6.1 16.1 37.9
2.1 6.1 17.4 6.1 16.1 37.9
28/297
8.5
11.0
DISCUSSION
Many authors have proposed endoscopic and clinical criteria for the identification of patients at high VOLUME 38, NO.4, 1992
1 2 3 4 5 6 Total
Rate of bleeding (%)
risk of bleeding from esophageal varices who have never bled. The results of the application of the proposed criteria have varied greatly from author to author and have rarely been confirmed in subsequent applications in different environments. 9 • 18• 19 Pagliaro et al.,2 reviewed 13 controlled trials on the efficacy of prophylactic sclerotherapy, applying metanalysis methods. They demonstrated that the selection of patients was one of the main factors in determining the results of the studies. In fact, the rate of hemorrhage among the control groups varied between 427
14 and 70%, while prophylactic sclerotherapy proved to be beneficial to the treated patients only if the actual bleeding risk for the untreated patients was greater than 40 to 50% in 1 year. In our series, the bleeding incidence calculated with the Kaplan-Meier method was 12.5% at 12 months, which is lower than reported elsewhere. Possible explanations are the relatively short follow-up (average 14 months) or the relatively low percentage of patients who were Child C class (64 of 297,21.5%). We would stress that in 75% of cases the source of bleeding, endoscopically determined, was represented by varices. These data are not very different from those of other authors 4.12. 19, 20 apart from the finding of a higher percentage of bleeding due to associated lesions, which confirms previous data from our group.16 During follow-up, 63 patients died (18.0% at 1 year and 23.8% at 2 years); half of them belonged to Child C class of liver decompensation. Mortality seemed to be directly related to the hemorrhagic event in one third of cases. Our prospective study confirms that bleeding due to portal hypertension, and particularly to the rupture of varices, is a dramatic event which, despite the great advances made in the treatment of the underlying liver condition, is attended by a very high mortality rate. If preventive measures are to be taken, patients at risk need to be much more accurately identified. In this study, even if only the highest risk classes are taken into account, the percentage of patients bleeding in 1 year is far from the 40%2 which is probably needed to justify an aggressive procedure like sclerotherapy. Moving from the simplest and probably most reproducible classification (that of Dagradi5) through the JRSPH classes to the NIEC scores, the highest risk of bleeding in 1 year ranges from 19.8% through 24.0% to 37.9%. However, this rise in risk is at the expense of the number of patients classified, for it decreases in absolute terms from 85 to 24 to 19 patients, respectively. If we examine the results of the three methods in relation to bleeding at 1 year and for the highest risk classes (Table 6), sensitivity is extremely low in all three cases, as is the predictive value of a positive test. Whichever of the three well-known classification methods for esophageal varices is adopted, there is no clear practical advantage for the population of patients with portal hypertension who have never bled. It is probably better, therefore, for reasons of reproducibility and ease of learning by novice endoscopists, to adopt the simplest type of classification based on variceal size (possibly taking into account red signs) or, alternatively, the simplified NIEC method, rather than more sophisticated ones. With all three classifications, the predictive bleeding values cast doubt on the usefulness of aggressive methods of bleeding prevention. In this type of patient, beta blockers are still 428
Table 6. Sensitivity, specificity, predictive value of positive and negative test and overall diagnostic accuracy in predicting the bleeding event of the highest risk classes calculated with the three classifications for esophageal varices Dagradi 5,6 (5th degree) Sensitivity Specificity Predictive value of positive test Predictive value of negative test Overall diagnostic accuracy
NIEC 12 JRSPH 10 (score < -1.14) (score> 40.0)
46.7% 75.5% 16.5%
16.7% 93.4% 20.8%
17.9% 94.8% 26.3%
93.2%
91.5%
91.7%
69.7%
86.2%
87.5%
probably the best available preventive treatment,20-23 but more controlled multicenter trials are necessary to assess their efficacy. The association of endoscopic evaluation with some independent method for the study of esophageal varices, like echo-Doppler, echoendoscopy, variceal pressure,24 the dynamic study of the evolution of varices,25 or some other invasive method will probably be needed before a better prediction system for variceal bleeding is available.
REFERENCES 1. D'Amico G, Morabito A, Pagliaro L, Marubini E, The Liver Study Group of V. Cervello Hospital. Survival and prognostic indicators in compensated and decompensated cirrhosis. Dig Dis Sci 1986;31:468-75. 2. Pagliaro L, Burroughs AK, Sorensen TIA, et al. Therapeutic controversies and randomised controlled trials (RCTs): prevention of bleeding and rebleeding in cirrhosis. Gastrointest Radiol 1989;2:71-4. 3. Palmer ED, Brick lB. Correlation between the severity of esophageal varices in portal cirrhosis and their propensity towards haemorrhage. Gastroenterology 1956;30:85-90. 4. Baker LA, Smith C, Liebermann G. The natural history of esophageal varices. A study of 115 cirrhotic patients in whom varices were diagnosed prior to bleeding. Am J Med 1959;26:228-37. 5. Dagradi AE. The natural history of esophageal varices in patients with alcoholic liver cirrhosis: an endoscopic and clinical study. Am J Med 1972;57:520-40. 6. Dagradi AE, Tempien SJ, Owens LK. Bleeding esophagogastric varices (an endoscopic study of 50 cases). Arch Surg 1966;92:944-7. 7. Lebrec D, De Fleury P, Rueff B, et al. Portal hypertension, size of esophageal varices, and risk of gastrointestinal bleeding in alcoholic cirrhosis. Gastroenterology 1980;70:1139-44. 8. Garsia-Tsao G, Groszmann RJ, Fisher RL, et al. Portal pressure, presence of gastroesophageal varices and variceal bleeding. Hepatology 1985;5:419-24. 9. Beppu K, Inokuchi K, Koyanagi N, et al. Prediction of variceal haemorrhage by esophageal endoscopy. Gastrointest Endosc 1981;4:4-5. 10. Japanese Research Society for Portal Hypertension. The general rules for recording endoscopic findings of esophageal varices. Jpn J Surg 1980;10:84-7. 11. Paquet KJ. Prophylactic endoscopic sclerosing treatment of the esophageal wall in varices-a prospective controlled randomised trial. Endoscopy 1982;14:4-5. 12. The North Italian Endoscopic Club for the Study and Treat-
GASTROINTESTINAL ENDOSCOPY
13. 14. 15. 16. 17. 18. 19.
ment of Esophageal Varices. Prediction of the first variceal haemorrhage in patients with cirrhosis of the liver and esophageal varices. A prospective multicenter study. N Engl J Med 1988;15:983-9. De Franchis R., Primignani M, Vazzoler MC, et al. Endoscopia digestiva ed ipertensione portale. Atti 6, Simposio Nazionale SlED, Rome, November 14-17,1990:50-60. Pugh RNH, Murray-Lyon 1M, Dawson JL, Piertoni MC, Williams R. Transection of the oesophagus for bleeding oesophagus varices. Br J Surg 1973;10:84-7. The Italian Liver Cirrhosis Project. Reliability of endoscopy in the assessment of variceal features. J Hepatol 1987;4:93-8. Rigo GP, Merighi A, Antonioli A, et al. Routine and emergency endoscopy in patients with esophageal varices. Ital J GastroenteroI1986;18:189-93. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457-81. Rigo GP, Merighi A, Chahin NJ, et al. Is emergency due to variceal bleeding predictable? Z Gastroenterol 1986;24(suppl 1):38. De Franchis R, The North-Italian Endoscopic Club. Prophylactic sclerotherapy (ST) in high risk cirrhotics selected by endo-
VOLUME 38, NO.4, 1992
20. 21.
22. 23.
24. 25.
scopic criteria: interim report of a multicenter randomised controlled trial. J HepatoI1988;7(suppl):41. Conn HO, Lindenmuth WW, May CJ, Rambsy GR. Prophylactic portacaval anastomosis. A tale of two studies. Medicine 1972;50:27-40. The Italian Multicenter Project For Propranolol in Prevention of Bleeding. Propranolol for prophylaxis of bleeding in cirrhotic patients with large varices: a multicenter, randomised clinical trial. Hepatology 1988;8:1-5. Ideo G, Bellati G, Fesce X, Grimoldi D. Nadolol can prevent the first gastrointestinal bleeding in cirrhotics: a prospective, randomised study. Hepatology 1988;1:6-9. Lebrec D, Poynard T, Bernuan J, et al. A randomised controlled study of propranolol for prevention of recurrent gastrointestinal bleeding in patients with cirrhosis: a final report. Hepatology 1984;4:355-8. Rigau J, Bosch J, Bordas JM, et al. Endoscopic measurement of variceal pressure in cirrhosis: correlation with portal pressure and variceal hemorrhage. Gastroenterology 1989;96:873-8. Rigo GP, Zanasi G, Codeluppi PL, et al. The class change velocity of oesophageal varices is predictive of haemorrhage risk in cirrhotic patients. Endoscopy 1988;20:19.
429
A prospective study of the ability of three endoscopic classifications to predict hemorrhage from esophageal varices G. P. Rigo, MD, A. Merighi, MD N. J. Chahin, MD, M. Mastronardi, MD P. L. Codeluppi, MD, A. Ferrari, MD C. Armocida, MD, G. Zanasi, MD A. Cristani, MD, G. Cioni, MD F. Manenti Modena, Italy
Hemorrhage from esophageal varices in cirrhotics is a frequent event with high mortality in spite of therapy. Preventive sclerotherapy seems to be beneficial only if the patient's bleeding risk is higher than 40 to 50% a year. A series of 320 patients with esophageal varices without previous bleeding was studied prospectively; the varices were classified according to three widely used endoscopic classifications. During follow-up (6 to 36 months, average 14 months), hemorrhage occurred in 49 patients (15.3%) of whom 30 (61.2%) bled from varices (8.2 and 11.0% at 12 and 24 months, respectively). At the same time intervals, mortality of the entire population studied was 18.0 and 23.8%, respectively, of which one third was directly due to hemorrhage. With all three classifications, the higher the degree of bleeding risk, the greater the actual percentage of hemorrhages recorded; however, it never reached 40% a year. In predicting the bleeding event, Dagradi's classification proved more sensitive than JRSPH or NIEC, but the latter classifications were more specific and assessed a higher predictive value for a positive test. Endoscopic observation probably needs integration with other methods if a reliable bleeding prediction is to be made. (Gastrointest Endosc 1992;38:425-429)
Patients suffering from portal hypertension frequently hemorrhage from esophageal varices. This is a dramatic event accompanied by a high incidence of mortality.1 Analysis of controlled trials on different types of prophylactic treatment (drugs, endoscopy, surgery) against the first episode of bleeding from esophageal varices has shown that it is necessary to select patients with a high risk of bleeding in order to get the best results. 2 Many authors have observed that patients with large varices are more at risk than those with small varices. 3- s Beppu et al} in a retrospective
Received June 20, 1991. For revision August 15, 1991. Accepted December 26, 1991. From Gastroenterology and Digestive Endoscopy and 3rd Department of Clinical Medicine, University of Modena, Modena, Italy. Reprint requests: Gianpiero Rigo, MD, Cattedia di Gastroenterologia ed EndoscopiaDigestiva, Policlinico, viadelPozzo 71, 41100 Modena, Italy. VOLUME 38, NO.4, 1992
study, applied the criteria of the Japanese Research Society for Portal Hypertension lO (JRSPH) to patients with esophageal varices and were able to calculate a score which identified high-risk patients (80 to 100%). Paquet,!1 in a controlled trial of prophylactic sclerotherapy, selected patients exhibiting large varices with black points or large varices associated with reduced prothrombin activity (30% of normal values). In 2 years of follow-up, 66% of the controls suffered from variceal bleeding. In 1988, The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices (NIEC)12 proposed a prognostic index for hemorrhage from esophageal varices taking into account the size of the varices, the presence of red wale markings, and Child's classification. Using this index, the authors were able to define six risk classes, in the last of which bleeding occurred in 68.9% of the patients in 1 year of follow425
was classified as non-variceal but due to associated lesions if there was actual bleeding from the lesion (erosions and ulcers, benign or malignant lesions of the esophagus, stomach and duodenum and acute hemorrhagic gastritis) or stigmata from the Forrest classification. Otherwise, the bleeding was classified as being of unidentified origin. 16 Mortality was considered to be related to bleeding only when it occurred within 6 weeks from the hemorrhage and was linked to it by a continuous sequence of events as evident from clinical records of the hospital where the patients had been admitted. Bleeding episodes and mortality were evaluated by means of the life table methodY
up. A simplified index was introduced more recently 13 with substantially identical results. The aim of our prospective study was 3-fold: to calculate the incidence of upper digestive tract hemorrhage due to esophageal varices and from other bleeding sites, to determine total mortality and that related to bleeding; and to evaluate the possibility of identifying high-risk patients using three classifications of esophageal varices in a population of patients with esophageal varices but without a history of previous bleeding. PATIENTS AND METHODS
All patients with esophageal varices diagnosed endoscopically in the Gastrointestinal Unit of the University of Modena between October 1, 1984 and February 28, 1986 were considered for the study. The criteria for exclusion were: (1) the existence of previous bleeding episodes from esophageal varices or of unidentified origin; (2) previous treatment (pharmacological, surgical, endoscopic) capable of modifying portal pressure or the risk of hemorrhage; and (3) associated diseases (such as cancer) with reduced life expectancy. All of the patients' data were computerized: they included all parameters (clinical, biochemical, and endoscopic) necessary for the calculation of Child's classification (modified by Pugh)14 and of the esophageal varices grading according to the classifications of Dagradi,5 JRSPH, and NIEC. Dagradi5 suggested five degrees depending on the size of the varices, the fifth degree being characterized by large varices with cherry red spotS. 5,6 The JRSPH classification proposed six classes of risk based not only on the evaluation of the four fundamental parameters (color, form, and extension of the varices and red signs) but also on the presence or absence of esophagitis. 10 The NIEC index also described six classes of risk but based only on three parameters: Child's class, size of varices, and the presence of red wale markings. 12 In order to ensure maximal homogeneity in the description of the esophageal varices, all endoscopists participating underwent a preliminary training program set up as part of the Italian Liver Cirrhosis Project. 15 All relevant clinical events were recorded 6, 12, and 24 months after initial classification. When direct examination was not possible, the patients or their physicians were contacted by phone at the intervals mentioned above. When a hemorrhage was reported during follow-up, the source of bleeding, as ascertained by emergency endoscopy, was classified as variceal when the following were observed: bleeding from the varices, white nipples on the varices, nonbleeding fifth-degree varices, and no other potential or apparent sources of bleeding in the upper digestive tract. It
RESULTS
From October 1, 1984 to February 28, 1986, 477 patients with esophageal varices were examined by the endoscopists of the unit. A total of 136 patients were excluded from the study: 52 patients because of previous bleeding episodes from varices or other unidentified lesions; 64 patients owing to previous treatment capable of modifying portal pressure and bleeding risk; and 20 patients because of liver hepatoma or other severe associated disorders. In 21 patients the data were insufficient for complete evaluation. The remaining 320 patients were considered to be suitable for purpose of this study. Table 1 gives the demographic data and diagnosis of the patients. Follow-up continued for 6 to 36 months, the average was 14 months. Hemorrhage: incidence and source of bleeding
During follow-up, 49 patients (15.3%) bled (Table 2). The bleeding was due to varices in 30 patients (61.2%) and to associated lesions in 10 patients (20.4%); in 9 patients (18.4%) the source of bleeding was unidentified. Of the bleeding episodes due to esophageal varices (Fig. 1), the incidence rose from 8.2% at 12 months to 11.0% at 24 months. Mortality: incidence and causes
During follow-up, 63 patients died; 18.0% at 12 months and 23.8% at 24 months (Table 2). Hemorrhage accounted for approximately one third of total deaths: bleeding was due to varices in nine patients, to associated lesions in two patients, and was of unidentified origin in eight patients (these patients died
Table 1. Data of patients at entry Diagnosis No. of patients
320 426
Sex
M/F
Age (yr)
231/89 54.8 ± 11
Child classification
Liver Secondary biliary Portal cirrhosis cirrhosis thrombosis
318
1
1
A
B
C
108 125 64
Not determined
23
GASTROINTESTINAL ENDOSCOPY
Table 2. Cumulative rate of bleeding and mortality
Bleeding Overall Variceal Associated lesions Unidentified origin Mortality Overall Due to bleeding
Total no. of patients
Rate (%) 12 months
24 months
49 30 10 9
12.5 8.2 2.2 2.1
20.9 11.0 4.5 5.4
63 19
18.0 5.4
23.8 6.6
Table 3. Cumulative percentage (Kaplan-Meier method) of variceal bleeding among 320 patients with portal hypertension and esophageal varices classified according to Dagradi 5 • 6 at entry Varices degree VI V2 V3 V4 V5 Total
No.ofpatients who bled/total
12 months
24 months
0/5 5/122 3/82 4/26 18/85
3.6 2.7 13.0 19.8
3.6 2.7 19.0 24.0
30/320
8.2
11.0
Rate of bleeding (%)
100
j ~
!
t . ~
Table 4. Cumulative percentage (Kaplan-Meier method) of patients bleeding among 320 patients with portal hypertension and esophageal varices classified according to JRSPH'o at entry
80
70
Risk class
JRSPH score
No. of patients Rate of bleeding (%) who bled/total 12 months 24 months
60
50
oJ----r----,---.--
-'~7--30"-
12 -'Tis--18"-'i,--2'" r-',
1 2 3 4 5 6
months
Figure 1. Cumulative percentage of patients free of variceal bleeding during the study (life table).
in other hospitals where no emergency endoscopic examination was performed).
>1.14 0.38 to 1.14 0.38 to 0.0 0.0 to -0.38 -0.38 to -1.14 <-1.14 Total
3/131 7/88 3/15 5/34 6/28 6/24
2.5 8.1 6.9 13.4 16.4 24.0
2.5 8.1 6.9 13.4 21.2 31.6
30/320
8.2
11.0
Table 5. Cumulative percentage (Kaplan-Meier method) of patients bleeding among 297 patients with portal hypertension and esophageal varices classified according to NIEC'2 at entry
Bleeding risk from esophageal varices
Risk
NIEC in-
No. of patients
The cumulative incidence of hemorrhagic episodes at 12 and 24 months varied, depending on the classification of risk adopted: according to the classification of Dagradi,5 it was markedly higher in patients with fourth- and fifth-degree varices than in those with second- and third-degree varices (Table 3). According to JRSPH, it increased progressively as the class of risk rose, reaching a maximum of 24.0 and 31.6%, respectively, in patients at maximum risk (Table 4). According to the NIEC index, it was significantly lower in patients with a risk factor <35.0 (classes I, II, III, and IV) than it was in those with a risk factor >35.0 (classes V and VI). The high rate of bleeding at 12 months (37.9%) in patients of the sixth class, with no further increase at 24 months, is particularly worth noting (Table 5).
class
dex
who bled/total
12 months
24 months
<20 20.0-25.0 25.1-30.0 30.1-35.0 35.1-40.0 >40
2/50 5/94 6/48 3/38 7/48 5/19
2.1 6.1 4.3 6.1 16.1 37.9
2.1 6.1 17.4 6.1 16.1 37.9
28/297
8.5
11.0
DISCUSSION
Many authors have proposed endoscopic and clinical criteria for the identification of patients at high VOLUME 38, NO.4, 1992
1 2 3 4 5 6 Total
Rate of bleeding (%)
risk of bleeding from esophageal varices who have never bled. The results of the application of the proposed criteria have varied greatly from author to author and have rarely been confirmed in subsequent applications in different environments. 9 • 18• 19 Pagliaro et al.,2 reviewed 13 controlled trials on the efficacy of prophylactic sclerotherapy, applying metanalysis methods. They demonstrated that the selection of patients was one of the main factors in determining the results of the studies. In fact, the rate of hemorrhage among the control groups varied between 427
14 and 70%, while prophylactic sclerotherapy proved to be beneficial to the treated patients only if the actual bleeding risk for the untreated patients was greater than 40 to 50% in 1 year. In our series, the bleeding incidence calculated with the Kaplan-Meier method was 12.5% at 12 months, which is lower than reported elsewhere. Possible explanations are the relatively short follow-up (average 14 months) or the relatively low percentage of patients who were Child C class (64 of 297,21.5%). We would stress that in 75% of cases the source of bleeding, endoscopically determined, was represented by varices. These data are not very different from those of other authors 4.12. 19, 20 apart from the finding of a higher percentage of bleeding due to associated lesions, which confirms previous data from our group.16 During follow-up, 63 patients died (18.0% at 1 year and 23.8% at 2 years); half of them belonged to Child C class of liver decompensation. Mortality seemed to be directly related to the hemorrhagic event in one third of cases. Our prospective study confirms that bleeding due to portal hypertension, and particularly to the rupture of varices, is a dramatic event which, despite the great advances made in the treatment of the underlying liver condition, is attended by a very high mortality rate. If preventive measures are to be taken, patients at risk need to be much more accurately identified. In this study, even if only the highest risk classes are taken into account, the percentage of patients bleeding in 1 year is far from the 40%2 which is probably needed to justify an aggressive procedure like sclerotherapy. Moving from the simplest and probably most reproducible classification (that of Dagradi5) through the JRSPH classes to the NIEC scores, the highest risk of bleeding in 1 year ranges from 19.8% through 24.0% to 37.9%. However, this rise in risk is at the expense of the number of patients classified, for it decreases in absolute terms from 85 to 24 to 19 patients, respectively. If we examine the results of the three methods in relation to bleeding at 1 year and for the highest risk classes (Table 6), sensitivity is extremely low in all three cases, as is the predictive value of a positive test. Whichever of the three well-known classification methods for esophageal varices is adopted, there is no clear practical advantage for the population of patients with portal hypertension who have never bled. It is probably better, therefore, for reasons of reproducibility and ease of learning by novice endoscopists, to adopt the simplest type of classification based on variceal size (possibly taking into account red signs) or, alternatively, the simplified NIEC method, rather than more sophisticated ones. With all three classifications, the predictive bleeding values cast doubt on the usefulness of aggressive methods of bleeding prevention. In this type of patient, beta blockers are still 428
Table 6. Sensitivity, specificity, predictive value of positive and negative test and overall diagnostic accuracy in predicting the bleeding event of the highest risk classes calculated with the three classifications for esophageal varices Dagradi 5,6 (5th degree) Sensitivity Specificity Predictive value of positive test Predictive value of negative test Overall diagnostic accuracy
NIEC 12 JRSPH 10 (score < -1.14) (score> 40.0)
46.7% 75.5% 16.5%
16.7% 93.4% 20.8%
17.9% 94.8% 26.3%
93.2%
91.5%
91.7%
69.7%
86.2%
87.5%
probably the best available preventive treatment,20-23 but more controlled multicenter trials are necessary to assess their efficacy. The association of endoscopic evaluation with some independent method for the study of esophageal varices, like echo-Doppler, echoendoscopy, variceal pressure,24 the dynamic study of the evolution of varices,25 or some other invasive method will probably be needed before a better prediction system for variceal bleeding is available.
REFERENCES 1. D'Amico G, Morabito A, Pagliaro L, Marubini E, The Liver Study Group of V. Cervello Hospital. Survival and prognostic indicators in compensated and decompensated cirrhosis. Dig Dis Sci 1986;31:468-75. 2. Pagliaro L, Burroughs AK, Sorensen TIA, et al. Therapeutic controversies and randomised controlled trials (RCTs): prevention of bleeding and rebleeding in cirrhosis. Gastrointest Radiol 1989;2:71-4. 3. Palmer ED, Brick lB. Correlation between the severity of esophageal varices in portal cirrhosis and their propensity towards haemorrhage. Gastroenterology 1956;30:85-90. 4. Baker LA, Smith C, Liebermann G. The natural history of esophageal varices. A study of 115 cirrhotic patients in whom varices were diagnosed prior to bleeding. Am J Med 1959;26:228-37. 5. Dagradi AE. The natural history of esophageal varices in patients with alcoholic liver cirrhosis: an endoscopic and clinical study. Am J Med 1972;57:520-40. 6. Dagradi AE, Tempien SJ, Owens LK. Bleeding esophagogastric varices (an endoscopic study of 50 cases). Arch Surg 1966;92:944-7. 7. Lebrec D, De Fleury P, Rueff B, et al. Portal hypertension, size of esophageal varices, and risk of gastrointestinal bleeding in alcoholic cirrhosis. Gastroenterology 1980;70:1139-44. 8. Garsia-Tsao G, Groszmann RJ, Fisher RL, et al. Portal pressure, presence of gastroesophageal varices and variceal bleeding. Hepatology 1985;5:419-24. 9. Beppu K, Inokuchi K, Koyanagi N, et al. Prediction of variceal haemorrhage by esophageal endoscopy. Gastrointest Endosc 1981;4:4-5. 10. Japanese Research Society for Portal Hypertension. The general rules for recording endoscopic findings of esophageal varices. Jpn J Surg 1980;10:84-7. 11. Paquet KJ. Prophylactic endoscopic sclerosing treatment of the esophageal wall in varices-a prospective controlled randomised trial. Endoscopy 1982;14:4-5. 12. The North Italian Endoscopic Club for the Study and Treat-
GASTROINTESTINAL ENDOSCOPY
13. 14. 15. 16. 17. 18. 19.
ment of Esophageal Varices. Prediction of the first variceal haemorrhage in patients with cirrhosis of the liver and esophageal varices. A prospective multicenter study. N Engl J Med 1988;15:983-9. De Franchis R., Primignani M, Vazzoler MC, et al. Endoscopia digestiva ed ipertensione portale. Atti 6, Simposio Nazionale SlED, Rome, November 14-17,1990:50-60. Pugh RNH, Murray-Lyon 1M, Dawson JL, Piertoni MC, Williams R. Transection of the oesophagus for bleeding oesophagus varices. Br J Surg 1973;10:84-7. The Italian Liver Cirrhosis Project. Reliability of endoscopy in the assessment of variceal features. J Hepatol 1987;4:93-8. Rigo GP, Merighi A, Antonioli A, et al. Routine and emergency endoscopy in patients with esophageal varices. Ital J GastroenteroI1986;18:189-93. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457-81. Rigo GP, Merighi A, Chahin NJ, et al. Is emergency due to variceal bleeding predictable? Z Gastroenterol 1986;24(suppl 1):38. De Franchis R, The North-Italian Endoscopic Club. Prophylactic sclerotherapy (ST) in high risk cirrhotics selected by endo-
VOLUME 38, NO.4, 1992
20. 21.
22. 23.
24. 25.
scopic criteria: interim report of a multicenter randomised controlled trial. J HepatoI1988;7(suppl):41. Conn HO, Lindenmuth WW, May CJ, Rambsy GR. Prophylactic portacaval anastomosis. A tale of two studies. Medicine 1972;50:27-40. The Italian Multicenter Project For Propranolol in Prevention of Bleeding. Propranolol for prophylaxis of bleeding in cirrhotic patients with large varices: a multicenter, randomised clinical trial. Hepatology 1988;8:1-5. Ideo G, Bellati G, Fesce X, Grimoldi D. Nadolol can prevent the first gastrointestinal bleeding in cirrhotics: a prospective, randomised study. Hepatology 1988;1:6-9. Lebrec D, Poynard T, Bernuan J, et al. A randomised controlled study of propranolol for prevention of recurrent gastrointestinal bleeding in patients with cirrhosis: a final report. Hepatology 1984;4:355-8. Rigau J, Bosch J, Bordas JM, et al. Endoscopic measurement of variceal pressure in cirrhosis: correlation with portal pressure and variceal hemorrhage. Gastroenterology 1989;96:873-8. Rigo GP, Zanasi G, Codeluppi PL, et al. The class change velocity of oesophageal varices is predictive of haemorrhage risk in cirrhotic patients. Endoscopy 1988;20:19.
429