Multivariate pathophysiological oesophagitis

Gut 1997; 40: 167-174

PAPERS

Multivariate analysis of pathophysiological factorsin reflux oesophagitis

G Cadiot, A Bruhat, D Rigaud, T Coste, A Vuagnat, Y Benyedder, T Vallot,D Le Guludec, M Mignon

Department ofHepato-Gastroenterology,Bichat-ClaudeBernard Hospital,Paris, FranceG CadiotA BruhatD RigaudY BenyedderT VallotM Mignon

Department ofHepato-Gastroenterology,Avicenne Hospital,Bobigny, FranceT Coste

INSERM U169,Villejuif, FranceA Vuagnat

Department ofNuclearMedicine,Bichat-ClaudeBernard Hospital,Paris, FranceD Le GuludecCorrespondence to:Dr G Cadiot,Service d'hepato-gastroenterologie, H6pitalBichat-Claude Bernard, 46,rue Henri-Huchard, 75877Paris Cedex 18, France.

Accepted for publication20 September 1996

AbstractBackground-Reflux oesophagitis is con-sidered a multifactorial disease, but therespective roles of the main factorsinvolved in its pathophysiology have notbeen clearly established.Ains-To attempt to assign these roles bymeans ofa multivariate logistic regressionanalysis of the main parameters asso-ciated with reflux oesophagitis.Patients-Eighty seven patients withgastro-oesophageal reflux disease werestudied: 41 without oesophagitis and 46with reflux oesophagitis grade 1 to 3.Methods-(l) Monovariate comparison ofpatients' characteristics and of para-meters derived from in hospital 24 houroesophageal pH monitoring, oesophagealmanometry, double isotope gastricemptying studies, and basal and penta-gastrin stimulated gastric acid and pepsinoutput determinations, between patientswith and without oesophagitis. (2) Multi-variate logistic regression analysis includ-ing the parameters significant in themonovariate analysis.Results-Among the 16 significant para-meters from monovariate analysis, threesignificant independent parameters wereidentified by multivariate logistic regres-sion analysis: number of refluxes lastingmore than five minutes, reflecting oeso-phageal acid clearance (p=0002); basallower oesophageal sphincter pressure(p=0.008); and peak acid output (p=0-012).These three parameters were not corre-lated with each other. The multivariatemodel was highly discriminant (correctclassification of 81.3% of the cases (95%confidence intervals 0-723, 0.903). Risk foroesophagitis increased as a function ofthetercile threshold values of the three para-meters. Odds ratios of the three para-meters for oesophagitis risk were similar,regardless ofwhether they were calculatedwhen the patients were compared as afunction of oesophagitis grade or thepresence or absence of oesophagitis.Conclusion-This multivariate approachadds evidence that impaired oesophagealacid clearance and hypotonic loweroesophageal sphincter are the two majorindependent pathophysiological factors ofoesophagitis, but also showed that the acid

secretion level is an independent patho-physiological factor.(Gut 1997; 40: 167-174)

Keywords: reflux oesophagitis, pathophysiology,multivariate analysis, oesophageal clearance, loweroesophageal sphincter, acid secretion.

When the antireflux barrier is impaired,permanently or transiently, gastro-oesophagealreflux may occur. The reasons why somepatients with gastro-oesophageal reflux disease(GORD) develop reflux oesophagitis are notfully understood.A" For many authors, refluxoesophagitis can be regarded mainly as a pri-mary motility disorder with disturbances ofoesophageal motility and clearance,5-9 loweroesophageal sphincter (LOS) incompe-tence,'0 11 and, for some authors, delay ingastric emptying. 2 However, many otherfactors could contribute to the creation ofoesophageal mucosal erosions such as thetoxicity of the refluxate,13 14 which depends onthe characteristics of the gastric juice, and theoesophageal mucosal resistance." 11 Further-more, the respective part played by each patho-physiological factor in the creation ofoesophageal erosions is unknown. Althoughreflux oesophagitis seems to be a multifactorialdisease, many studies have only consideredeach of its potential pathophysiological factors,separately. Few studies have considered theassociation of several pathophysiologicalfactors of oesophagitis.17-23 None used a multi-variate model including the main factors differ-entiating between GORD patients with andwithout oesophagitis.The aim of this study was to evaluate the

respective roles of the main factors involved inthe pathophysiology of reflux oesophagitis bymeans of multivariate logistic regression analy-sis. The studied parameters were patients'characteristics and parameters derived from 24hour oesophageal pH monitoring, standardoesophageal manometry, double isotopegastric emptying studies, and basal and penta-gastrin stimulated acid and pepsin secretiondeterminations. A monovariate comparison ofthese parameters was first carried out in twogroups of GORD patients, with or withoutreflux oesophagitis, then the significant para-meters were included in the multivariate analy-sis. Odds ratios of the identified independentparameters for oesophagitis risk were de-termined.

167

on February 9, 2022 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.167 on 1 F

ebruary 1997. Dow

nloaded from

Cadiot, Bruhat, Rigaud, Coste, Vuagnat, Benyedder, Vallot, Le Guludec, Mignon

Methods

PATIENTS

Outpatients suffering from GORD were pro-spectively recruited by two of us (TC, MM).Patients who had undergone upper digestiveendoscopy performed by three of us (GC, TC,TV) and in hospital 24 hour oesophageal pHmonitoring could be included in the study whenthe following inclusion and exclusion criteriawere fulfilled. The inclusion criteria were: (1)typical symptoms ofGORD - that is, heartburn,regurgitation or both, for more than six months- or atypical GORD related symptoms, such aschronic bronchopulmonary symptoms, chroniclaryngitis or non-cardiac chest pain, and (2)excessive oesophageal acid exposure on inhospital 24 hour oesophageal pH monitoring -that is, >4-2% of reflux time at pH <4 on 24hour recording or >6/3% in the upright position,or >1 2% in the supine position,24 or all three.Exclusion criteria were: (1) previous gastricor oesophageal surgery; (2) diabetes mellitus;(3) progressive systemic sclerosis, Raynaud'ssyndrome and other vascular collagen diseases;(4) neurological disorders; (5) Zollinger-Ellisonsyndrome; and (6) complicated oesophagitis -that is, oesophageal ulcer or stenosis andBarrett's oesophagus (>3 cm of columnar celllined oesophagus). Patients who gave theirinformed consent and underwent oesophagealmanometry, double isotope gastric emptyingstudy, and basal and pentagastrin stimulatedgastric acid and pepsin output determinationswere included. The study was conducted inaccordance with the Helsinki Declaration.

STUDY DESIGN

EndoscopyUpper digestive tract endoscopy was performedroutinely without sedation using Olympus endo-scopes. The absence of oesophagitis was definedas the absence of erosions (no oesophagitisgroup). Oesophagitis was graded as follows:grade 1: erosion(s) involving only a single longi-tudinal oesophageal fold; grade 2: non-circum-ferential mutiple erosions involving more thanone longitudinal fold with or without conflu-ence; grade 3: circumferential erosions.25 Hiatalhernia, defined as a distance between the dia-phragmatic hiatus and the gastro-oesophagealjunction >2 cm, was diagnosed endoscopically.Helicobacter pylori status was not assessed.

In hospital 24 hour oesophagealpH monitoringAll studies were performed with a glass pHelectrode (Ingold 10-4403022, diameter 3-8mm) connected to a portable digital recorder(Digitrapper, Synectics Medical). Before eachrecording, the pH electrode was standardisedagainst buffer solutions at pH 7 0, 4 0, and 1I0.The pH probe was inserted through the noseand its tip was positioned 5 cm above thegastro-oesophageal junction, identified pHmetrically and not manometrically. All thepatients were prescribed the same standardiseddiet: 2085 kcal/24 h, 15-8% protein, 36.6% fat,

and 47T6% carbohydrates with meals at7 00-7 30, 12 00-12 30, 18 30-19 00. Patientsrecorded the time of food or fluid consumptionand posture changes on a diary card. Data wereanalysed on an IBM PC (software: Esogram/Gastrogram 9822, version 5 09, Gastrosoft).The following variables were recorded duringthe 24 hour period and during the upright andsupine periods: number of refluxes lasting >5minutes (NR5) at pH <4, percentage of refluxtime at pH <4, number of refluxes at pH <4,and duration of the longest reflux at pH <4.

Oesophageal manometryAfter an overnight fast, oesophageal mano-metry was performed using a 4 catheter waterperfused tube (external diameter, 5 mm)connected to an Arndorfer hydrocapillarypump. The distal side openings on the cath-eters were staggered radially at 900 intervalsand were spaced 5 cm apart. The station pullthrough technique was used to determine thebasal LOS pressure (LOSP) at the pressureinversion point on each catheter. Intervalsbetween steps were 20-30 seconds. The meanof four values obtained in each patient wascalculated. The amplitudes of at least 10oesophageal waves were measured 5 cm and 10cm above the LOS after swallowing 5 ml ofwater at 20 to 30 second intervals and themeans of these wave amplitudes were calcu-lated. The percentage of completely propa-gated oesophageal waves between 10 cm and5 cm above the LOS was determined.

Double isotope gastric emptying studyGastric emptying of both solid and liquidphases of a meal was studied as previouslydescribed.26 Briefly, after an overnight fast, adouble labelled test meal of 450 kcal was eatenin less than 15 minutes at 11 00. The mealcontained 21 g of protein, 23 g of fat, and40 g of carbohydrate, in a final volume of 400ml. The liquid phase of the meal was labelledby 100 ,uCi of ["'In]DTPA diluted with tepidwater (220 ml) and the solid phase with 1 mCiof 99mTc-ovalbumin mixed into eggs. The 237keV (±20%) peak for gamma emissions wasselected as the indium 'window' and the 140keV (+ 1 0%) peak as the technetium 'window'.An Acti-gamma camera (CGR, Sopha-Medical, Paris, France) was used, connected toa Digital Equipment (Sismis III, Sopha-Medical). The region corresponding to thestomach was determined on both anterior andposterior images. Successive one minuteanterior and one minute posterior images werecollected at 10 minute intervals for 200minutes, with the subject in an uprightposition. Results for both solids and liquids areexpressed as time elapsed until 50% emptying(emptying half time) had been achieved and aslag time (min) for emptying solids.

Gastric acid and pepsin output determinationsTests were performed after an overnight fast aspreviously described.27 Briefly, a nasogastric

168

on February 9, 2022 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.167 on 1 F

ebruary 1997. Dow

nloaded from

Pathophysiology of reflux oesophagitis

tube was placed in the stomach and its positionchecked by the water recovery test, while thepatient lay in a semirecumbent position. Thegastric juice was collected by gentle manualaspiration and pooled in 15 minute samplesduring the basal state for 75 minutes and inresponse to pentagastrin (6 pug/kg intramuscu-larly) for 60 minutes. The first 15 minutesample was discarded. Subsequent 15 minutesamples were stored at 4°C immediately aftercollection. The volume of each sample wasmeasured. Acidity was determined by titrationup to pH 7-0 and proteolytic activity (pepsin)by an automated method using human haemo-globin as the substrate.28 Basal hourly acid andpepsin outputs were calculated. Peak acid(PAO) and pepsin outputs in response topentagastrin were calculated as twice the sumof the two consecutive 15 minute sampleshaving the highest acid or pepsin outputs.Gastric acid outputs are expressed as mmol/hand pepsin outputs as pepsin units/h.

identified using an ascending selection pro-cedure (Proc Logistic, SAS 6-08, Cary, NC,USA). The final model included only thoseparameters with a p value <0.05. Thresholdvalues of the independent significant para-meters were determined according to theterciles of each parameter. Odds ratios of thesethreshold values were calculated for oeso-phagitis risk. The selected parameters werealso included in a multivariate logistic re-gression analysis comparing the patients as afunction of the grade of oesophagitis.

ResultsEighty seven GORD patients were studied: 41had no oesophageal erosion (no oesophagitisgroup); 46 had oesophageal erosions (oeso-phagitis group). In this second group, 10patients (22%) had grade 1 oesophagitis, 30patients (65%) had grade 2 oesophagitis, andsix patients (1/3%) had grade 3 oesophagitis.

STATISTICS

Results are expressed as medians and ranges.Monovariate comparisons between the twogroups of patients, with and without oeso-phagitis, were made using the Kruskal-Wallistest or the Mann-Whitney U test for quan-titative parameters and the x2 test for quali-tative parameters with Yates's correction when-ever relevant. Correlations between parameterswere measured by the Spearman rank corre-lation coefficient. Statistical significance wasset at p<0 05. Parameters significant accordingto the monovariate analysis between patientswith and without oesophagitis were included ina multivariate logistic regression analysis.Independent significant parameters were

TABLE I Charactenrstics of the two groups ofGORD patients

No oesophagitis OesophagitisParameter (n=41) (n=46) p Value

Age (y) median (range) 47 (24-71) 58 (26-81) 0-025Women/men 22/19 13/33 0-015BMI (kg/m2) median (range) 24-0 (16-3-31-0) 25-9 (20-7-35-4) 0-005Hiatal hernia 18 34 0-004

BMI: body mass index: weight/height2. p Values for age and BMI were determined using Mann-Whitney U test, and for sex and hiatal hernia with the x2 test.

TABLE II OesophagealpH monitoring data in the two groups ofGORD patients

No oesophagitis (n=41) Oesophagitis (n=46)Parameter median (range) median (range) p Value*

NR5Total 4-0 (0-0-19-0) 8-0 (0 0-30-0) 0-0002Upright 3-0 (0-0-15-0) 4-0 (0-0-17-0) 0-039Supine 1-0 (0-0-16-0) 2-3 (0-0-26-0) 0-130

Reflux time (/)Total 6-9 (1-4-44-0) 12-2 (36-51-5) 0-0009Upright 9 9 (1-3-27-2) 12-9 (3-3-68-0) 0-030Supine 6-2 (0-1-62-8) 8-0 (0-2-65-4) 0-240

No of reflux episodesTotal 46-0 (7-0-157-0) 64-0 (13-0-216-0) 0-030Upright 31-0 (2-0-121-0) 49-5 (8-0-150-0) 0-010Supine 8-0 (2-0-42-0) 13-0 (2-0-68-0) 0-071

Longest reflux (min)Total 31-0 (4-0-208-0) 31-0 (2-0-128-0) 0 440Upright 13-0 (1-0-71-0) 19-5 (2-0-128-0) 0 053Supine 34-0 (1-0-208-0) 20-5 (1-0-204-0) 0-430

*Mann-Whimey U test. Total: during the 24 hour monitoring period. Upright: during theupright period monitoring. Supine: during the supine monitoring period.

MONOVARIATE ANALYSIS

Patients' characteristics (Table I)Compared with patients without oesophagitis,patients with oesophagitis were significantlyolder (p=0 025), had a significantly higherbody mass index (p=0 005), were more oftenmen (p=0-0 15), and more of them had a hiatalhernia (p=0 004).

In hospital 24 hour oesophagealpH monitoring(Table II)Among oesophageal 24 hour pH monitoringdata, the most relevant factor was NR5; itsmedian was two times higher in the group withoesophagitis than in the group without(p=0 0002). NR5 also differed significantly bygroup when patients were compared accordingto their grade of oesophagitis (p=0 002) (Fig 1),and there was a significant linear correlationbetween NR5 and the grade of oesophagitis(r=0 41, p<0 001). The difference betweenNR5 values for patients with hiatal hernia and

030 r

20 H

et

qs*CN

z10

0

0

0

8_ o

oo

OoDOOC_ liq?

0 00

080

80

0

00

00

0

co0

-9o-00

0 1 2 3Grade of oesophagitis

Figure 1: Individual numbers of refluxes lasting >5 minutes(NRS) during the 24 hour oesophagealpH recording as afunction of the grade of oesophagitis.25 Bars indicatemedians. p=0-002 (Kruskal-Wallis test).

169

on February 9, 2022 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.167 on 1 F

ebruary 1997. Dow

nloaded from

Cadiot, Bruhat, Rigaud, Coste, Vuagnat, Benyedder, Vallot, Le Guludec, Mignon

TABLE m Oesophageal and gastnc motility data in the two groups ofGORD patients

No oesophagitis (n=41) Oesophagitis (n=46)Parameter median (range) median (range) p Value*

LOSP (mm Hg) 18-0 (4-0-33 0) 11-8 (4 5-25 0) 0-0008Ampl 5 (mm Hg) 87-0 (36-0-183-0) 73-0 (20-0-195-0) 0-150Ampl 10 (mm Hg) 79-0 (37-0-162-0) 63-0 (10-0-161-0) 0 050Propagation (O/6) 100-0 (80-0-100-0) 100-0 (80-0-100-0) 0-480T,/2 solids (min) 120-0 (60-0-195-0) 105-0 (45-0-180-0) 0-280T1,2 liquids (min) 75-0 (15-0-150-0) 75-0 (30-0-165-0) 0-670Lag time (min) 18-2 (0-0-137-0) 16-1 (0-0-150-4) 0-840

*Mann-Whitney U test. Ampi 5 and 10: oesophageal wave amplitude measured 5 and 10 cmabove the LOS. T,,2: gastric emptying half time. Lag time: lag time for gastric emptying of solids.

30

I

E

E

-

Cicnm

m

20

10

o

Figure XsphincteoesophaWallis t

thosestatistipercerhigheroesop]

median amplitude of oesophageal waves

measured 10 cm above LOS was 20-3% lowerin patients with oesophagitis (p=005). Thepercentage of completely propagated oeso-

phageal waves did not differ significantly be-tween the two groups. In 12 patients withoesophagitis and 15 patients without oeso-

phagitis, 90% or less of the oesophageal waves

were completely propagated (NS).

Double isotope gastric emptying study (Table III)Median gastric emptying halftime of solids and

o liquias ana lag time tor emptying OI soilas wereo comparable for the two groups.

80

0

Gastric acid and pepsin output determinations°° (Table IV)

(a80o 8 °° Median PAO 25-9% higher in oesophagitis

ETe 80 patients compared with patients without8 oesophagitis (p=0 007). When comparing the0g,° patients according to their grade of oeso-

-o phagitis, PAO tended towards significance

(p=0-061) (data not shown). There was a slight°8° ° but significant linear correlation between PAO

and the grade of oesophagitis (r=-028,p<0010). Eight (17A4%) oesophagitis patients

0 1 2 3 had PAO values above the mean PAO+2 SDGrade of oesophagitis of the non oesophagitis group - that is, 51

2: Individual values of the basal lower oesophageal mmol/h. The percentage of patients with PAO?rpressure (LOSP) as afunction of the grade of >51 mmolh increased significantly with theigitis.2Bars indicate medians. p=0005 (Kruskal- grade of oesophagitis (p=0016) that is,

test). 24%/, 0%, 20%, and 33% of the patients withno oesophagitis, grade 1, grade 2, and grade 3oesophagitis, respectively. Median basal acid

without hiatal hernia did not reach output, basal pepsin output, and peak pepsinical significance (p=006). The median output were higher in the patients with oeso-

itage of total reflux time (at pH <4) was phagitis than in those without oesophagitis,r in the oesophagitis group than in the no +122-7% (p=0 033), +85-5% (p=0 009), andhagitis group (+76-8%) (p=0 0009). The +35-2% (p=0 020), respectively.

percentage of total reflux time was significantlycorrelated with NR5 (r=0 90, p<0O0001).Significantly higher numbers of reflux episodeswere recorded in the oesophagitis group. The

pH monitoring data also differed significantlybetween the two groups in the upright positionbut not in the supine position.

Oesophageal manometry (Table III)Basal median LOSP in patients with oeso-

phagitis was 34 4% lower than in patientswithout oesophagitis (p=0 0008). Further-more, LOSP also differed significantly betweengroups when patients were compared accord-ing to their grade of oesophagitis (p=0 005)(Fig 2). There was a weak but significant linearcorrelation between basal LOSP and the gradeof oesophagitis (r=-038, p<0O001). The

TABLE Iv Gastric acid and pepsin outputs in the two groups ofGORD patients

No oesophagitis (n=41) Oesophagitis (n=46)Output median (range) median (range) p Value*

Basal acid (mmol/h) 2-2 (00-25 7) 4-9 (00-19-8) 0 033Peak acid (mmol/h) 27-0 (84-55-3) 34 0 (58-75-9) 0 007Basal pepsin (PU/h) 200-0 (00-1450-0) 371-0 (00-1205-0) 0 009Peak pepsin (PU/h) 790-0 (1450-2270-0) 1068-0 (1800-2906-0) 0-020

*Mann-Whitney U test. PU: pepsin unit.

MULTIVARIATE LOGISTIC REGRESSION ANALYSIS

The 16 significant parameters identified bymonovariate analysis were included in the multi-variate logistic regression analysis. These factorswere: age, sex, body mass index, presence ofhiatal hernia, NR5 during the 24 hour periodand during the upright period, percentage ofreflux time during the 24 hour period andduring the upright period, number of refluxepisodes during the 24 hour period and duringthe upright period, basal LOSP, amplitude ofcontraction waves measured 10 cm above LOS,

TABLE v Multivariate logistic regression analysis betweenGORD patients with or without oesophagitis

Odds Odds ratioParameters Estimate ratio* 95% CI p Value

NR5 0-1539 1-166 1-057, 1-287 0-002LOSP -0-1167 0-890 0-816, 0 970 0-008PAO 00508 1-052 1-011, 1-095 0-012Intercept -0-8944 0 409 0-060, 2-767 0-359

The multivariate analysis included the 16 significantparameters identified by monovariate comparison between thetwo groups of GORD patients. *: with a one point increment.The odds ratio of each parameter was adjusted to the other twoparameters. CI=confidence intervals. The model correctlyclassified the patients (no oesophagitis or oesophagitis) in81.-3% of the cases (95% CI: 0-723, 0 903).

170

on February 9, 2022 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.167 on 1 F

ebruary 1997. Dow

nloaded from

Pathophysiology of reflux oesophagitis

TABLE VI Spearman correlation analysis between the threeindependent parameters

Parameters r p Value

NR5/LOSP -0-164 >0-10NR5/PAO 0 109 >0 10LOSP/PAO -0-147 >0-10

basal acid output, PAO, basal pepsin output,and peak pepsin output. The multivariatelogistic regression analysis identified threeindependent factors that contribute significantlyto differentiating between patients with andwithout oesophagitis - that is, NR5 (p=0 002),basal LOSP (p=0-008), and PAO (p=0-012)(Table V). Body mass index and male sex hadadjusted p values tending towards significance,p=0-062 and p=0067, respectively. Otherparameters had adjusted p values above 0-14.The model, -0-8944+0-1539XNR5-0-1167

XLOSP+0 0508xPAO, correctly classifiedpatients as having oesophagitis or not in 81-3%of the cases (95% confidence intervals: 0-723,0-903). This model fits our data very well(Hosmer and Lemeshow goodness of fit test)with a non-significant p value of 0-98.The absence of correlation between NR5,

LOSP, and PAO was determined by Spearmanrank correlation analysis (Table VI).

ODDS RATIOS OF NR5, LOSP, AND PAO FOR

OESOPHAGlTIS RISK

Odds ratios, with each parameter beingadjusted at constant level to the other twoparameters, were calculated with a one pointincrement for NR5, LOSP, and PAO, and are

given in Table V.The impact on the risk for oesophagitis of

different thresholds of these three independentparameters, determined according to terciles,are indicated in Table VII. Compared with theparameter thresholds associated with thelowest oesophagitis risk - that is, NR5 <4/24 h,LOSP .18 mm Hg, and PAO <25 mmol/h,respective odds ratios of the other parameterthresholds were significantly higher, with theexception of the parameter threshold 4 .NR5<8/24 h (p=0-081). Because of the smallnumbers of patients, the determination ofadjusted odds ratios for the combination of theparameter thresholds was not possible.Odds ratios of NR5, basal LOSP, and PAO,

calculated when the patients were divided intofour groups as a function of oesophagitis grade(no oesophagitis, grade 1, grade 2, and grade 3)were very similar to the values calculated whenthe patients were divided into two groups, withand without oesophagitis (Table V). Oddsratios (95% CI) calculated as a function of theoesophagitis grade were: 1- 127 (1-048, 1-213)for NR5, 0-887 (0-822, 0-957) for basal LOSP,and 1-047 (1-015, 1-081) for PAO.

DiscussionThe reasons why some GORD patientsdevelop reflux oesophagitis remain unclear. Asmany parameters differentiate between GORDpatients with and without oesophagitis, it is

TABLE via Risk for oesophagitis according to differentthresholds ofNR5, LOSP, and PAO

Odds Odds ratioParameter thresholds ratio 95% CI p Value

NR5 <4/24 h 1-00 - -4 <NR5 <8/24 h 2-76 0-88, 8-59 0-081NR5 28/24 h 6-86 2-15, 21-91 0-001LOSP >18 mm Hg 1-00 - -11-8 LOSP <18 mm Hg 4-58 1-44,14-57 0-010LOSP < 11 8 mm Hg 6-32 2-11, 18-97 0-001PAO <25 mmol/h 1-00 - -

25 <PAO <36 mmol/h 3-12 1-04, 9-33 0-042PAO>36mmolVh 4 50 1-46, 13-89 0 009

Thresholds were determined according to the terciles of eachparameter.

generally accepted that the pathogenesis ofreflux oesophagitis is multifactorial.lA In thisstudy, most of the known parameters involvedin the pathophysiology of reflux oesophagitiswere investigated in each patient. To identifywhich factors independently differentiate thesepatients and to increase our understanding ofthe pathophysiology of reflux oesophagitis, weundertook a multivariate logistic regressionanalysis of the main known parameters thatdistinguished between GORD patients withand without oesophagitis after monovariateanalysis.The multivariate analysis showed that three

independent parameters were significantly as-sociated with oesophagitis - that is, NR5,LOSP, and PAO. These three parameters werenot correlated with each other (Table VI). Theodds ratio of each parameter for oesophagitisrisk was calculated and the cumulative effect ofthese parameters on oesophagitis risk wasestablished using the multivariate model. Ourmodel was highly discriminant, as it correctlyclassified the patients into the two groups -with and without oesophagitis - in 81 3% ofthe cases. Furthermore, its goodness of fit withour data was very high. These findings seemrelevant because the oesophagitis risk asso-ciated with different thresholds ofNR5, LOSP,and PAO increased with the increase in thethreshold values of the three parameters (TableVII). This finding suggests a continuous effectof each parameter on oesophagitis risk.Furthermore, when the patients were com-pared as a function of the oesophagitis grade,NR5 and LOSP differed significantly (Figs 1and 2), but PAO only tended towards signifi-cance (p=0-061). A significant linear corre-lation was also found between the values ofNR5, LOSP, and PAO and the grade of oeso-phagitis. Finally, odds ratios calculated as afunction of the oesophagitis grade (fourgroups) were very similar to those calculated asa function of the absence or presence ofoesophagitis (two groups).Do these statistical findings have patho-

physiological significance? Many studies haveshown, as we did, that NR5 was higher inGORD patients with reflux oesophagitis thanin those without oesophagitis,17 23 24 29-33 andalso that NR5 increased with the oesophagitisgrade.32 34 NR5 is a marker of both oeso-phageal acid clearance and the amount of acidthat refluxes, and was highly significantlycorrelated with the total percentage of reflux

171

on February 9, 2022 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.167 on 1 F

ebruary 1997. Dow

nloaded from

Cadiot, Bruhat, Rigaud, Coste, Vuagnat, Benyedder, Vallot, Le Guludec, Mignon

time 7 24 29 (r=@090 in our study). Although thepercentage of total reflux time was highly signi-ficant after monovariate analysis (Table II), itwas not identified as such by the multivariateanalysis, probably because this parameter,unlike NR5, does not give information aboutoesophageal acid clearance. This also suggeststhat if acid exposure was an important factorof oesophagitis, independent from oesophagealacid clearance, the percentage of total refluxtime would have been identified by themultivariate analysis. However, considering thehigh correlation between the percentage oftotal reflux time and NR5, we cannot excludethat the multivariate analysis underestimatedthe role of acid exposure. Using differentmethods, it has been shown that oesophagealacid clearance is impaired in reflux oeso-phagitis and is considered an importantpathophysiological factor in reflux oeso-phagitis.5 8 17 30 35-43 Therefore, our study,using a multivariate approach, confirmedthat impairment of oesophageal acid clear-ance is the main determinant of reflux oeso-phagitis.Recent data showed the role of hiatal hernia

in clearance impairment in patients with oeso-phagitis.43-5 However, the presence of a hiatalhernia, which was highly associated with refluxoesophagitis according to our monovariateanalysis, was not identified as such in themultivariate model. NR5 had a non-significanttendency to be higher in patients with hiatalhernia than in patients without hiatal herniaand the only studied parameter reflectingoesophageal acid clearance (amplitude of theoesophageal waves measured 10 cm aboveLOS) that differed significantly between thetwo groups of GORD patients after mono-variate analysis was not identified in the multi-variate model. Therefore, our study does notallow us to speculate on the mechanisms ofoesophageal acid clearance impairment inpatients with oesophagitis.LOS is a major factor in gastro-oesophageal

competence which, in turn, is a majordeterminant of reflux oesophagitis.l1 6 How-ever, the pathophysiological significance of thebasal LOSP value remains uncertain. Indeed,although it has been clearly determined thatvery low basal LOSP values are implicated inthe pathophysiology of severe oesophagitis,many GORD patients with moderate oeso-phagitis have normal basal LOSPvalues.'0 1 19 32 46 47 Most of the gastro-oesophageal refluxes occur during transientLOS relaxations in GORD patients.10 11 48Furthermore, there is considerable overlap ofbasal LOSP values of individual GORDpatients with and without oesophagitis andnormal controls. Our study showed that twothirds of the patients had basal LOSP valuesover 11 8 mm Hg (normal values). However,oesophagitis risk was clearly increased whenbasal LOSP values were below 18 mm Hg.This effect was much more significant forvalues below 1 1-8 mm Hg. Are these findingsa cause or a consequence of oeso-phagitis?l 7 9 37 49 FThe results of our study donot allow us to resolve this question.

PAO reflects the acid secretory status of thepatients. Although conflicting results havebeen reported concerning the relation betweengastric acid secretion and reflux oeso-phagitis,' 3 14 19 21 31 50 our multivariate modelshowed that PAO was also an independentfactor associated with reflux oesophagitis.When comparing GORD patients with andwithout oesophagitis, PAO differed signifi-cantly (Table IV). Furthermore, there was adose dependent increase in odds ratios as afunction of the PAO values (Table VII). How-ever, the correlation between PAO and thegrade of oesophagitis was poor (r=0-28,p<00 10) and PAO only tended towards signi-ficance when patients were compared as afunction of their grade of oesophagitis(p=0-061). Nevertheless, the percentage ofpatients with very high PAO values increasedsignificantly with the grade of oesophagitis.Many studies have shown, after monovariatecomparison, that PAO, maximal acid output,and the percentage of patients with acid hyper-secretion were not significantly higher inGORD patients with oesophagitis comparedwith those without oesophagitis.13 14 21 There-fore, many authors do not think that thecomposition of gastric juice in itself is a factorcontributing to oesophagitis in GORDpatients. However, many arguments suggestthat the level of gastric acid secretion, which isa determinant of the composition of therefluxate, influences oesophagitis risk. Firstly,one of the most important arguments is theefficacy of gastric acid antisecretory drugs,which do not modify motility parameters, topromote oesophagitis healing.37 51 Their effectdepends on the level of acid reduction.50 53Secondly, in Zollinger-Ellison syndrome, achronic condition of acid hypersecretion,reflux oesophagitis is very common,54 55 theabsence of significant abnormalities of oeso-phageal motility and LOS clearly suggests thatthe composition of the gastric juice in thiscontext is sufficient to provoke oesophagitis.56That the PAO level was also identified as anindependent factor does not mean that patientswith oesophagitis are acid hypersecretors com-pared with GORD patients without oeso-phagitis; it merely indicates that, within anindividual, there is an independent increasingrisk of oesophagitis relative to the PAO value.The mechanisms of increased PAO were notstudied, because the pertinent factors were notassessed, in particular, Helicobacterpylori status.To the best of our knowledge, studies ongastric acid parameters and GORD relatedmotility parameters did not establish clearrelations between these parameters. 14 Ourstudy showed that PAO was not correlated withNR5 - that is, oesophageal acid clearance andbasal LOS tone.

All possible oesophagitis risk factors werenot evaluated, especially oesophageal mucosalresistance.'9 16 Oesophageal mucosal resis-tance, combined with oesophageal clearance, isprobably one of the main factors of defencelimiting the toxicity of the refluxate. Itassociates pre-epithelial, epithelial, and post-epithelial mechanisms that currently are not

172

on February 9, 2022 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.167 on 1 F

ebruary 1997. Dow

nloaded from

Pathophysiology of reflux oesophagitis

well known.'5 16 To our knowledge, no simpletest evaluating oesophageal mucosal resistance,suitable for clinical studies, is available, and therespective part of oesophageal mucosal resis-tance among all pathophysiological factors ofreflux oesophagitis has not been studied.Although oesophageal mucosal resistance andother possible oesophagitis risk factors, such astransient LOS relaxations,'0 11 48 were notevaluated, our multivariate model includingNR5, LOSP, and PAO was highly discriminantbecause it correctly classified patients as havingoesophagitis or not, in 81 3% ofthe cases (95%CI, 0-723, 0 903). This high discriminationrate obtained with the multivariate modelsuggests that these three independent para-meters reflect the main pathophysiologicalfactors of reflux oesophagitis (grades 1 to 3).

In conclusion, this multivariate analysis con-firms that reflux oesophagitis is multifactorial,with impairment of oesophageal acid clearanceand basal LOSP. However, our study alsoshowed that the level of gastric acid secretionwas an independent parameter. These findingsare concordant with the current knowledge ofthe pathophysiology of oesophagitis and needto be tested on an independent population.Furthermore, they also support the currentmedical therapeutic strategies - that is, gastricacid antisecretory drugs and prokinetic agents.

1 Dodds WJ, Hogan WI, Helm JF, Dent J. Pathogenesis ofreflux esophagitis. Gastroenterology 1981; 81: 376-94.

2 Richter JE, Castell DO. Gastroesophageal reflux: patho-genesis, diagnosis and therapy. Ann Intern Med 1982; 97:93-103.

3 De Caestecker JS, Heading RC. The pathophysiology ofreflux. In: TPJ Hennessy, A Cuschieri, JR Bennett, eds.Reflux oesophagitis. London: Butterworths, 1989: 1-36.

4 Timmer R, Breumelhof R, Nadorp JHSM, Smout AJPM.Recent advances in the pathophysiology of gastro-oesophageal reflux disease. Eur J Gastroenterol Hepatol1993; 5: 485-91.

5 Kahrilas PJ. Esophageal motor activity and acid clearance.Gastroenterol Clin North Am 1990; 19: 537-50.

6 Katzka DA, DiMarino AJ. Pathophysiology of gastro-esophageal reflux disease: LES incompetence and eso-phageal clearance. In: DO Castell, ed. The esophagus.Boston: Little, Brown, 1992: 449-61.

7 Castell DO. Gastroesophageal reflux disease is a motilitydisorder. In: C Scarpignato, Ed. Advances in drug therapyof gastroesophageal reflux disease. Basel: Karger. FrontGastrointest Res 1992; 20: 11-6.

8 Eriksen CA, Sadek SA, Cranford C, Sutton D, Kennedy N,Cuschieri A. Reflux oesophagitis and oesophagitis transit:evidence for a primary oesophageal motor disorder. Gut1988; 29: 448-52.

9 Sonnenberg A, Lepsien G, Muller-Lissner SA, Koelz HR,Siewart JR, Blum AL. When is esophagitis healed? Eso-phageal endoscopy, histology and fimction before andafter cimetidine treatment. Dig Dis Sci 1982; 27:297-302.

10 Dodds WJ, DentJ, Hogan WJ, HelmJF, Hauser R,Patel GK,et al. Mechanisms of gastroesophageal reflux in patientswith reflux oesophagitis. N Engl J Med 1982; 307:1547-52.

11 Dent J, Holloway RH, Toonli J, Dodds WJ. Mechanisms oflower oesophageal sphincter incompetence in patientswith symptomatic gastro-oesophageal reflux. Gut 1988;29: 1020-8.

12 McCallum RW, Berkowitz DM, Levin E. Gastric emptyingin patients with gastroesophageal reflux. Gastroenterology1980; 80: 285-91.

13 Dubois A. Pathophysiology of gastroesophageal refluxdisease: role of gastric factors. In: DO Casteil, ed. Theesophagus. Boston: Little, Brown, 1992: 479-91.

14 Cadiot G, Sekera E, Mignon M. Gastric secretion ingastroesophageal reflux disease. In: C Scarpignato, JPGalmiche, eds. Functional evaluation in esophageal disease.Basel: Karger. Front Gastrointest Res 1994; 22: 209-22.

15 Goldstein JL, Schlesinger PK, Mozwecz HL, Layden TJ.Esophageal mucosal resistance: a factor in esophagitis.GastroenterolClinNorthAm 1990; 19: 565-86.

16 Orlando RC. Pathophysiology of gastroesophageal refluxdisease: esophageal epithelial resistance. In: DO Castell,ed. The esophagus. Boston: Little, Brown, 1992: 463-78.

17 Little AG, DeMeester TR, Kirchner PT, O'Sullivan GC,Skinner DB. Pathogenesis of esophagitis in patients withgastroesophageal reflux. Surgery 1980; 88: 101-7.

18 Baldi F, Corinaldesi R, Ferrarini F, Stanghellini V,Miglioli M, Barbara L. Gastric secretion and emptyingof liquids in reflux esophagitis. Dig Dis Sci 1981; 26:886-9.

19 BruleydesVarannes S, TouchaisJY, WeberJ, DesechalliersJP,Sauger F, Denis P, et al. Oesophagite de reflux: r6lesrespectifs des facteurs pathogeniques oesophagiens etgastriques. Gastroenterol Clin Biol 1986; 10: 662-8.

20 Barlow AP, DeMeester TR, Ball CS, Eypasch EP. Thesignificance of the gastric secretory state in gastro-esophageal reflux disease. Arch Surg 1989; 124: 937-40.

21 Hirschowitz BI. A critical analysis, with appropriatecontrols, of gastric acid and pepsin secretion in clinicalesophagitis. Gastroenterology 1991; 101: 1149-58.

22 Zhu H, Pace F, Sangaletti 0, Bianchi Porro G. Gastric acidsecretion and pattern of gastroesophageal reflux inpatients with esophagitis and concomitant duodenalulcer. A multivariate analysis of pathogenetic factors.Scandj3 Gastroenterol 1993; 28: 387-92.

23 Sontag SJ, Schnell TG, Miller TQ, Nemchausky B,Serlovsky R, O'Connell S, et al. The importance of hiatalhernia in reflux esophagitis compared with lower eso-phageal sphincter pressure or smoking. Jf Clin Gastro-enterol 1991; 13: 628-43.

24 DeMeester TR, Wang CI, Wernly JA, Pellegrini CA,Little AG, Klementschitsch P, et al. Technique, indi-cations, and clinical use of 24 hour esophageal pH moni-toring. J Thorac Cardiovasc Surg 1980; 79: 656-70.

25 Ollyo JB, Fontollet CH, Monnier PH. Savary's newendoscopic grading of reflux oesophagitis: a simple,reproducible, logical, complete and useful classification.Gastroenterology 1990; 89: A100.

26 Rigaud D, Bedig G, Merrouche M, Vulpillat M, Bonfils S,Apfelbaum M. Delayed gastric emptying in anorexianervosa is improved by completion of a renutritionprogram. Dig Dis Sci 1988; 33: 919-25.

27 Vatier J, Poitevin C, Mignon M. Sialic acid content andproteolytic activity in gastric juice in humans. Anapproach for appreciating mucus glycoprotein erosion.Dig Dis Sci 1988; 33: 144-51.

28 Vatier J, Sauvage M, Lewin M, Bonfils S. Approche d'unedetermination automatique de l'activite proteolytique dusuc gastrique dans les conditions de la vitesse initiale dela reaction. Biol Gastroenterol 1975; 8: 107-15.

29 Schlesinger PK, Donahue PE, Schmid B, Layden TJ.Limitations of 24-hour intraesophageal pH monitoring inthe hospital setting. Gastroenterology 1985; 89: 797-804.

30 Pujol A, Grande L, Ros E, Pera C. Utility of inpatient24-hour intraesophageal pH monitoring in diagnosis ofgastroesophageal reflux. Dig Dis Sci 1988; 33: 1134-40.

31 Sekera E, Cadiot G, Poitevin C, VallotT, VatierJ, Mignon M.Secretion gastrique de pepsine dans le reflux gastro-oesophagien complique ou non d'oesophagite peptique.Gastroenterol Clin Biol 1992; 16: 141-7.

32 Bruley des Varannes S, Ravenbakht-Charifi M, Cloarec D,Pujol P, Simon J, Galmiche JP. Endobrachyoesophage etreflux gastro-oesophagien acide. Enregistrements pH-metriques etages et etude manometrique. GastroenterolClin Biol 1992; 16: 406-12.

33 Parrilla P, Ortiz A, Martinez de Haro LF, Aguayo JL,Ramirez P. Evaluation of the magnitude of gastro-oesophageal reflux in Barrett's oesophagus. Gut 1990; 31:964-7.

34 Fiorucci S, Santucci L, Chiucchiu S, Morelli A. Gastricacidity and gastroesophageal reflux patterns in patientswith esophagitis. Gastroenterology 1992; 103: 855-61.

35 Stanciu C, Bennett JR. Oesophageal acid clearing: onefactor in the production of reflux oesophagitis. Gut 1974;15: 852-7.

36 Williams D, Thompson DG, Marples M, Heggie L,O'Hanrahan T, Mani V, et al. Identification of anabnormal esophageal clearance response to intraluminaldistension in patients with esophagitis. Gastroenterology1992; 103: 943-53.

37 Singh P, Adamopoulos A, Taylor RH, Colin-Jones DG.Oesophageal motor function before and after healing ofoesophagitis. Gut 1992; 33: 1590-6.

38 Ferguson MK, Ryan JW, Little AG, Skinner DB. Eso-phageal emptying and acid neutralization in patients withsymptoms of esophageal reflux. Ann Surg 1985; 201:728-35.

39 DeCaestecker JS, Blackwell JN, Pryde A, Heading RC.Daytime gastro-oesophageal reflux is important in oeso-phagitis. Gut 1987; 28: 519-26.

40 Kahrilas PJ, Dodds WI, Hogan WI. Effect of peristalticdysfunction on esophageal volume clearance. Gastro-enterology 1988; 94: 73-80.

41 Tolin RD, Maimud LS, Reilley J, Fisher RS. Esophagealscintigraphy to quantitate esophageal transit (quantitationof esophageal transit). Gastroenteroloy 1979; 76: 1402-8.

42 Orr WC, Robinson MG, Johnson LF. Acid clearance duringsleep in the pathogenesis of reflux esophagitis. Dig Dis Sci1981; 26: 423-7.

43 Lin S, Ke M, Xu J, Kahrilas PJ. Impaired esophagealemptying in reflux disease. Am J Gastroenterol 1994; 89:1003-6.

44 Mittal RK, Lange RC, McCallum RW. Identification andmechanism of delayed esophageal acid clearance insubjects with hiatus hernia. Gastroenterology 1987; 92:130-5.

45 Sloan S, Kahrilas PJ. Impairment of esophageal emptyingwith hiatal hernia. Gastroenterology 1991; 100: 596-605.

46 Kahrilas PJ, Dodds WJ,Hogan WI,KernM, Arndorfer RC,Reece A. Esophageal peristaltic dysfunction in pepticesophagitis. Gastroenterology 1986; 91: 897-904.

173

on February 9, 2022 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.167 on 1 F

ebruary 1997. Dow

nloaded from

Cadiot, Bnuhat, Rigaud, Coste, Vuagnat, Benyedder, Vallot, Le Guludec, Mignon

47 Csendes A, Maluenda F, Braghetto I, Csendes P,Henriquez A, Quesada MS. Location of the loweroesophageal sphincter and the squamous columnarmucosal junction in 109 healthy controls and 778 patientswith different degrees of endoscopic oesophagitis. Gut1993;34:21-7.

48 Mittal RK, McCallum RW. Characteristics and frequencyof transient relaxations of the lower esophageal sphincterin patients with reflux esophagitis. Gastroenterology 1988;95: 593-9.

49 Williams D, Thompson DG, Heggie L, O'Hanrahan T,Bancewicz J. Esophageal clearance function following treat-ment of esophagitis. Gastroenteroly 1994; 106: 108-16.

50 Collen MJ, Lewis JH, Benjamin SB. Gastric acid hyper-secretion in refractory gastroesophageal reflux disease.Gastroenterology 1990; 98: 654-61.

51 Bell NJV, Hunt RH. Role of gastric acid suppression in thetreatment of gastro-oesophageal reflux disease. Gut 1992;33: 118-24.

52 Johansson KE, Tibbling L. Gastric secretion and refluxpattern in reflux oesophagitis before and duringranitidine treatment. Scand Gastroenterol 1986; 21:487-92.

53 Jansen JBMJ, Lamers CBHW. High doses of ranitidine inpatients with reflux oesophagitis. Scand Gastroenterol1990; 25 (suppl 178): 42-6.

54 Miller IS, Vinayek R, Frucht H, Gardner JD, Jensen RT,Maton PN. Reflux esophagitis in patients withZollinger-Ellison syndrome. Gastroenterology 1990; 98:341-6.

55 Cadiot G, Helie C, Vallot T, Marmuse JP, Cosnes J, RicheA, et al. Oesophagite stenosante au cours du syndrome deZollinger-Ellison. Gastroenterol Clin Biol 1994; 18:1018-20.

56 Strader DB, Benjamin SB, Orbuch M, Lubensky TA,Gibril F, Weber C, et al. Esophageal function andoccurrence of Barrett's esophagus in Zollinger-Ellisonsyndrome. Digestion 1995; 56: 347-56.

174

on February 9, 2022 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.40.2.167 on 1 F

ebruary 1997. Dow

nloaded from