||Improving Diagnostic Accuracy of Upper GI Endoscopy in Patients with GERD
||Mohammad Farivar, M.D.
||Caritas Norwood Hospital and Boston University Medical School
One Lenox Street
Norwood, MA 02062
the United States each year billions of dollars are spent on diagnosis and
treatment of GERD.
the "sine qua non" for GERD is present in less than half of patients
with GERD. EGD, the most expensive and the most used test in evaluating patients
with GERD is diagnostic for evidence of reflux esophagitis in less than 50%
of cases. Our observations indicate that examination of the larynx during
EGD (LEGD), and if necessary, histopathologic and morphometric study of properly
performed endoscopic esophageal biopsy, significantly enhances diagnostic
yield of EGD in patients with GERD.
avoid further diagnostic studies and to save cost, endoscopists should try
to become familiar with endoscopic evidence of RL, and perform to LEGD instead
of EGD, and if in doubt, obtain proper biopsy studies (A). In addition to
RL, patulous LES and HH are significant endoscopic findings indicating the
presence of reflux and more severe diseases. This study indicates that dyspepsia
is also another manifestation of GE reflux.
Running Title: Farivar,
EGD in GERD
Keywords: GERD, EGD, RL,
LEGD, HH, LESC
GER: Gastroesophageal Reflux
GERD: Gastroesophageal Reflux Disease
H & R: Heartburn and Regurgitation
EGD: Esophago-Gastro Duodenoscopy (upper
RL: Reflux Laryngitis
LESC: Lower Esophageal Sphincter Chalasia (incompetent
or patulous sphincter)
HH: Hiatal Hernia
BZT: Basal Zone Thickness
NCCP: Non-Cardiac Chest Pain
has been the focus of many recent review articles (6,7,8,9,10,11). Gastroesophageal
reflux (GER) is common among the population of the United States (2).
is estimated that heartburn, the most common symptom of GER occurs daily
in 7% of the US population (2) and is the major reason for the consumption
of antacids (3,4,5). Gastroesophageal reflux disease (GERD) is the most common
reason for performing upper GI endoscopies.
can be misunderstood for the following reasons: Symptoms may vary; heartburn
and regurgitation, occurs in only about 50% of patients (2); The natural
history of disease, the frequency of its spontaneous remission, complications
if untreated, and the goals of treatment are not fully understood; and there
is no diagnostic standard for GERD. The two most commonly performed studies
are: ambulatory 24hr pH Monitoring and upper GI endoscopy (EGD); 24hr pH
Monitoring is the most sensitive test for reflux and is considered the "Gold
Standard" by some. This test has several drawbacks including: incorrect
placement, tip migration, behavioral modifications during the study, lack
of reproducibility, inability to accept or tolerate the probe, and false
negative results in a subgroup of patients suffering from duodenogastroesophageal
reflux (14-24). Indeed, 23-29% of patients with erosive esophagitis have
normal acid exposure studies (31).
is the diagnostic procedure of choice by most gastroenterologists when evaluating
reflux patients, due to direct visualization of the esophageal mucosa, the
safety, ease and availability of the procedure, the immediate results, the
accuracy of findings, and the good quality of permanent records with video
endoscopy when esophageal mucosal disruption is present (95% specificity
in non-immunocompromised patients). However, endoscopy has the following
draw backs of expense, technical skill, inter-observer variations, subjectivity
in reporting results due to the lack of accepted uniformity in grading mucosal
damage and esophagitis, and insufficient knowledge of clinical significance
of Hiatus Hernia, Patulous LES, and Reflux Laryngitis (RL). Above all 39-85%
of patients with significant GER related symptoms undergoing EGD do not have
gross mucosal injuries and therefore have negative EGD (32-41). In Johnson
L., Fetal study only 54% of patients with reflux documented by pH studies
had endoscopic mucosal injury (40).
addition of esophageal biopsy performed during endoscopy can enhance endoscopic
diagnostic yield in patients with chronic reflux, when basal zone thickness
(BZT) is more than 1/6th and papillary height (PH) is more than 50% of total
mucosa thickness (42,43). Intraepithelial
polymorphonuclear cells and eosinophils when present more specifically indicate
inflammation (44,45,46). The mucosal changes of chronic esophagitis are patchy
(47) and most biopsy studies have been performed on multiple, well-oriented,
relatively large, suction biopsy specimens, containing mucosa and submucosa
value of commonly obtained endoscopic, small, poorly oriented, pinch biopsy
samples containing only epithelium are considered limited (53-62). Similar
to endoscopy, there is a lack of uniformity in diagnostic criteria and subjectivity
in its assessment. Morphometric studies are rarely conducted, hence leading
to a significant number of false positive and false negative reports. We
have shown that in order to increase the diagnostic value of biopsies, endoscopists
must obtain 3-4 biopsies proximal to 2cm from squamocolumnar junction. Pathologists
need be familiar with the pathology of chronic esophagitis and to perform
morphometry on all specimens.
on this discussion, there is little doubt that the tests available to identify
symptomatic reflux are not very sensitive in practice today. The challenge
is to develop other tests or more uniform criteria on the tests we already
have, in order to increase our diagnostic capabilities in patients with GERD.
GI endoscopy has grown to be a major component of our health care system,
due to a rapid increase in the volume and total cost in recent years. The
effectiveness of EGD depends on the skill of the endoscopist and the impact
of relevant information gained by endoscopy on patient’s diagnosis, management,
association between GER and posterior laryngitis was first described by Cherry
and Margolis in 1968(61) and is considered virtually pathognomonic for reflux.
However, endoscopists have paid very little, if any, attention to the laryngeal
area during EGD (62-72).
undertook a prospective study in order to specifically determine the prevalence
and significance of RL in patients undergoing EGD, and to see whether its
presence enhanced diagnostic accuracy of EGD in patients with GERD.
variations of endoscopy for esophagitis was eliminated by using modified
variations of endoscopic biopsy was eliminated by morphometry when no polys
or eosinophils were present. Inter-observer variation of reflux laryngitis
can be reduced by a learning curve.
MATERTALS & METHODS
study involved one hundred seventy six consecutive adult patients (18-91
years old) undergoing EGD. Of these, there were 70 male (average age of 54
years) and there were 106 female (average age 49 years). 18 of the patients
admitted to drinking alcohol and 32 smoked tobacco (8 did both) and 5 had
a variety of environmental allergies.
symptoms were divided into two categories: dyspeptic (upper abdominal pain,
nausea, vomiting, bloating); and GERD related. The
GERD symptoms included esophageal reflux (heartburn, regurgitation or dysphagia),
non-cardiac chest pain (NCCP), pulmonary symptoms (asthma, recurrent aspiration
pneumonia, etc) and ENT symptoms (hoarseness, chronic dry cough, chronic
sore throat, globus sensation).
endoscopy (EGD) was performed after obtaining informed consent using the
Olympus video-endoscopy system (GIF100). Patients received intravenous meperidine
and/or midazolam for conscious sedation. During endoscopy, esophageal findings
were classified as abnormal when mucosal disruption was present. To further
classify findings and eliminate the inter-observer variations, a modified
Savary-Miller criteria was used for grading esophagitis (73). Grade
I was considered when there were single or multiple erosions or linear
erythema in a single longitudinal fold of the esophagus. Grade II was
considered when single or multiple erosions were present on multiple folds.
We considered Grade III in cases with confluent mucosal denudation
and friability of the distal esophagus. Finally, the presence of inflammatory
stricture, deep ulcer, or Barrett's esophagitis was classified as Grade
addition to the above gradings, each patient was checked for presence of
hiatal hernia, without over inflating the lower esophagus. No attempt was
made to measure the axial size of the hernia or to classify it as a reducing
or non-reducing type. The status of the lower esophageal sphincter (LBS)
was carefully observed for the presence or absence of the patulous chalatic
sphincter. After completion of EGD, and as the scope was being withdrawn,
the larynx was observed (laryngo-EGD=LEGD) for evidence of reflux laryngitis
(RL). Reflux laryngitis was classified as “mild” when erythema was present
in the posterior third of the vocal cords and arytenoid area; as “moderate" when
diffuse posterior erythema, edema and piled up mucosa was seen in the inter-arytenoid
area; and as “severe” when ulceration or granulation was seen in the are
of the posterior vocal cords (62).
the endoscopy, 3 or 4 pinch biopsies were taken from grossly normal appearing
esophageal mucosa, 2-4 cm from the squamous-columnar junction of the esophagus.
The specimens were put on filter paper and placed in 10% formalin for fixation
and subsequent Hematoxylin and Eosin staining. The requisitions were labeled “abdominal
pain rule out chronic esophagitis", in order to avoid
observer bias (no true clinical history or endoscopic findings were revealed
to the pathologists). The biopsies were studied and signed out by any one
of the six members of the pathology department. Then, for the purpose of
this study, one pathologist (JK) reviewed all biopsy specimens for histopathology
and morphometric measurements without knowledge of patient history, prior
histopathological diagnosis or endoscopic findings. The
BZT and PH were measured according to previously published protocol.
We chose the criteria of BZT>15% and PH>50% of the total mucosal thickness as evidence for chronic reflux
be performed in 155 (88%) of the cases, 55% of those were measured on two
or more fragments. Overall histopathological diagnoses without morphometry
were misleading (false positive or false negative) in 27% of the patients.
Reflux Laryngitis: (Fig.
1) 136 of 176 patients (77%) had some degree of RL (more than 95% were mild
to moderate). Among 136 patients, 15(11%) had ENT symptoms, 31 (23%) had
positive endoscopy (erosive esophagitis), 113 (83%) had positive biopsies,
63 (46%) had hiatal hernia and 58 (43%) had heartburn and regurgitation (H&R).
As shown in figure
2, a significant number of patients with RL have more H&R (P<0.0l)
and positive biopsy (P<0.05) as compared to patients without RL. Furthermore,
RL was present in 94% of patients with LESC, 87% with hiatal hernia, 84%
of patients with endoscopic esophagitis, 67% of patients with stricture and
50% of patients with Barrett's esophagus. RL had the highest prevalence in
patients with patulous lower esophageal sphincter and the lowest occurrence
in patients with Barrett's (Fig.3).
Hiatal Hernia: (HH)
was present in 11 of 40 cases (27%) of patients without RL and 63 of 136
patients (46%) with RL. The prevalence of HH was greatest in patients with
endoscopic esophagitis, especially if associated with stricture and/or Barrett's,
and lowest (23%) in patients with only H&R. 63 of 73 (87%)of patients
with HH had RL. While the prevalence of HH was less than 50% in all other
symptomatic subgroups of patients with GERD, it was significantly higher
in patients with dysphagia (Fig.4).
Positive Endoscopy (Erosive-Ulcerative
Esophagitis): In 37 of patients with erosive esophagitis, 31(84%) had RL,
32(87%) had positive biopsy and 30(81%) had HH. As far as symptoms are concerned
17(46%) had heartburn, 11(32%) had chest pain, 5 had dysphagia, and 5 had
ENT or pulmonary symptoms. In 8 patients (22%) of the symptoms were not related
to the chest, and of these 4 patients had upper abdominal pain with or without
nausea and vomiting and 4 patients had occult GI bleeding (Fig.5).
In contrast, in 13
patients with normal endoscopy and severe heartburn as the only symptom 12
of the 13(92%) had RL, 11(85%) had positive biopsy and only 3(23%) had HH.
This data indicates that in patients with reflux induced erosive esophagitis,
HH is significantly more common and perhaps a contributing factor. In 5 patients
with Schatzki's Ring, 2 had RL (40%), 5 had a positive biopsy (100%) and
5 had HH (100%). Even though this is a small sample number, this data points
to a possible role for reflux and HH in the pathogenesis of Schatzki's Ring.
In 12 patients with stricture, 8(67%) had RL, 8(67%) had positive biopsies,
6(50%) had erosive esophagitis, 2(17%) had Barrett's esophagitis, and 12
(100%) had HH. In 10 patients with Barrett's, 5(50%) had RL and 9(90%) had
Sphincter Chalasia (LESC): Of the 176 patients undergoing EGD, 34(19%) had patulous lower esophageal
sphincter. Among the patients with LESC 9(26%) had erosive esophagitis.
32(94%) patients had RL and 28(82%) had positive biopsy. All those with
positive biopsy had RL. In 23(68%) of the patients in addition to RL, LESC
was the only other endoscopic finding. 19 of these patients (83%) had a
variety of GERD related symptoms, 59% had H&R, 21% had dyspepsia, and
24% had chronic sore throat or a persistent dry cough felt to be originating
in the laryngeal area. In addition, 14% had chest pain and 21% had dyspeptic
symptoms (Fig.6). In this group of patients with LESC, 12 had esophageal
manometry. LES pressure (normal values 10-25 mmHg) was 6-10 mmHg in 3 patients
and 5 mmHg or less in 9 patients. In 8 other patients with manometrically
proven LESP of 5 mmHg or less, LESC was not appreciated during endoscopy.
Therefore, one cannot always diagnose LESC based only on endoscopy.
and Regurgitation (H&R): In a total of 77 patients with severe heartburn (with or without
regurgitation) and other GERD related symptoms, 26(34%) had positive endoscopy,
69(90%)had RL, and 60(83%) had positive biopsy (fig.7). In this population
31 patients (40%) had HH and 23(30%) had LESC. When this group was subdivided
and data on 21 patients with only heartburn and regurgitation was examined
there were no significant changes. Among these patients 8(38%) had erosive
esophagitis, 19(90%) had RL and 18(86%) had positive biopsy. When the prevalence
of H&R was looked at in different endoscopic subgroups of GERD, its
occurrence was less than 50% in all, except in patients with LESC which
approached 60% (fig.8).
Chest Pain (NCCP):
In 42 patients with NCCP suspected of esophageal origin (9), EGD was positive
in 13(31%), RL was found in 33(79%) and biopsy was positive in 30(64%).
In 26 patients with negative endoscopy, 11 had LESC and 6 had HH (total
66%). When patients with only chest pain were analyzed, of the 18, EGD
was positive in 5(28%), RL was found in 9(50%), and biopsy was positive
in 7(39%). In 4 of the 18(22%) both RL and biopsies were negative and in
6(33%) both were positive. Therefore, the lack of RL and negative biopsy
effectively rules out reflux induced NCCP.
patients were complaining of various degrees of dysphagia. Among these, 12(25%)
had HH with erosive esophagitis and stricture, 5(11%) had HH and Schatzki’s
Ring, 3(6%) had CA of the esophagus, 3(6%) had Barrett's with stricture,
4 had only HH, and 6 had esophageal spasm during endoscopy. In total 30 patients
had HH (64%), 29 had RL (62%) and 33(80%) had positive biopsy. In 19 patients
whose only symptom was dysphagia to solids, 14(74%) had positive EGD, 15(79%)
had positive biopsy, 8(42%) had RL and 11(58%) had HH. Only 12 (26%) of patients
with dysphagia had H&R.
ENT & Pulmonary
Symptoms: Among 40
patients with ENT and pulmonary symptoms, 18(45%) had heartburn, 32(80%)
had RL, 34(85%) had positive biopsy, 9(22.5%) had positive EGD. In 31 patients
with normal endoscopy 12(39%) had HH and 6(19%) had LESC. In 18 patients
with only ENT symptoms EGD was positive in 3(17%), RL positive in 15(83%)
and biopsy positive in 15(83%). Of these patients, 7(39%) had HH and 2(11%)
had LESC (Fig.10). No significant changes were observed from primary group.
Among patients with ENT and pulmonary symptoms, one used tobacco, one drank
alcohol and none had environmental allergies. As shown in Fig. 1, only
11% of patients with RL had ENT and pulmonary symptoms.
19 patients with dyspeptic symptoms (upper abdominal pain, nausea, vomiting,
early satiety, bloating, and negative upper abdominal ultrasound), EGD revealed
gastritis, duodenitis, and gastric ulcer in 7 patients (37%) and two patients
had erosive esophagitis. Also, 13(68%) had RL and 13(68%) had positive biopsy.
Both positive biopsy and RL were found in 10(53%) of the patients, thereby
indicating that at least in 1/2 of the patients with dyspepsia and negative
endoscopy, reflux may he the cause of symptoms. This group of patients with
dyspepsia responded to antireflux treatment.
27 patients with no symptoms suspicious for GERD, 9 had no RL. Among them,
6 of 9 patients (66%) had gastritis, 2 patients had duodenitis, 1 patient
had gastric ulcer, 1 patient had normal endoscopy, 1 patient had iron deficiency
anemia with Grade II esophagitis, and 2 patients had abdominal pain, nausea,
studies have shown that acid can reach the level of the larynx (Jacob and
Weiner). The posterior larynx is the most exposed area due to positional
and gravitational forces (Ward Delahanty). It has been shown that gastric
content and a pH of four is capable of markedly damaging the laryngeal mucosa,
suggesting that Pepsin rather than acid may be the primary injurious agent.
Furthermore, intermittent and relatively infrequent laryngeal exposure to
GER can produce significant peptic upper aerodigestive tract injury (Kaufman).
this study the prevalence of RL was highest (94%) in patients with severe
heartburn and patulous LES, and lowest in patients suffering from comp1ications
of GER (as low as 50% in patients with Barrett's). In patients with complicated
RE the prevalence of HH was paradoxically higher as compared with RL. Ambulatory
pH studies have shown that the presence of HH especially when larger than
2 cm increases the frequency of GER and prolongs acid clearance time from
the esophagus, thereby increasing esophageal contact time with injurious
gastroduodenal content, and causing more severe forms of esophagitis in patients
with compromised LES (27, 28, 29, 30, 32), In this group of patients the
overall prevalence of HH was less than 50% in patients with GER, but approached
80-100% in patients with severe erosive esophagitis, peptic stricture, Barrett's
esophagitis arid Schatzki's Ring.
has been shown that acid and Pepsin together are more damaging to the esophageal
epithelium than either one alone. The addition of bile salt increases the
damage as well (38). Erosive esophagitis (69, 70, 71) and Barrett's esophagitis
(66, 68) have been found to be associated with more injurious alkaline-acid
reflux. Alkaline refluxers also suffer more from regurgitation, asthma, and
pulmonary fibrosis, and have less heartburn (72). Based on the above discussion,
possible explanations for the reduced prevalence of RL in patients with complicated
RE are as follows: a) the presence
of HH retains materials in the distal esophagus and prevents them from reaching
the level of the larynx, therefore, in addition to reducing the prevalence
of RL, by increasing contact time in the distal esophagus, it causes more
injury and subsequent complications in the distal esophagus; b) Some patients
with complicated RE, in addition to the presence of HH and incompetent LES,
are suffering from reflux by a more injurious mixture of acid, Pepsin, and
bile acids (duodenogastroesophageal reflux). However, as this mixture travels
up the esophagus and reaches the laryngeal level it loses its concentration
of Hydrogen, Pepsin and bile acids, and the volume is reduced. Also, as it
mixes with saliva it further reduces the damaging effects to the larynx by
increasing its pH. We have not observed RL in patients who suffer from pure
bile reflux, raising he possibility that bile acids alone are not injurious
to the laryngeal mucosa.
on its high prevalence it seems that in patients with GER, RL is the first
endoscopic sign to appear and the last to disappear after antireflux surgery
(Deveney) or the use of Omeprazole (Kemel Wilson). Similar to erosive esophagitis,
laryngeal symptoms do not respond adequately to H2 blockers (Kaufman & MF).
Endoscopic, as well as treatment data, suggest that laryngeal tissue is very
sensitive to acid and in order to heal RL and relieve symptoms when present,
laryngeal acid exposure must be totally eliminated. In our experience, long
term treatment with Omeprazole (20-40 mg. daily for at least 6 months) is
necessary. In some patients with dry cough due to reflux, the addition of
corticosteroid derivative inhaler (i.e. Vanceril 4-6 puffs bid for the first
few weeks of treatment) to Omeprazole has added benefits in relieving symptoms.
to our findings, others have also shown by videolaryngoscopy that RL is a
predictor of histological abnormalities in patients with reflux (05). In
this study mild RL was a common endoscopic finding. H&R and biopsy evidence
of chronic esophagitis were significantly more common in patients with RL.
Studies have shown that 6.5% to 51% of patients with reflux have ENT symptoms
(N65, 267). In our study group, 11% of patients with GER had upper aerodigestive
tract complaints, indicating that posterior laryngeal erythema is an indication
of reflux as well as a common endoscopic finding in patients with GER. However
RL is not the cause of various laryngeal symptoms.
esophagitis is present in less than 50% of patients with GERD except in those
with dysphagia to solids when the endoscopic yield of positive findings is
74%. In other GERD related conditions this was low (17-34%). Endoscopic evidence
of erosive esophagitis is even less common in patients with ENT symptoms
(17%). This prevalence has been reported to be 0-30% (Frazer, 72N).
this study the most common reason for endoscopy was for the evaluation of
GER related complaints (85% of upper GI endoscopy). Considering the high
cost of upper endoscopy, clinicians must be aware of its low yield when recommending
it and the endoscopists must pay attention to every possible abnormality
that may indicate the presence of reflux in order to avoid further diagnostic
studies, save cost, reduce discomfort and morbidity related to the procedure.
Becoming familiar with RL and looking for it during endoscopy can improve
the poor diagnostic capability of EGD in patients with GERD. This study has
shown that when the larynx is examined for evidence of RL, the diagnostic
yield or EGD increases from 34, 28 and 17% to 90, 67, and 80% in patients
with heartburn and regurgitation, GERD related chest pain, and ENT symptoms
is not considered among the etiologic factors in patients with dyspepsia
(84). Dyspepsia is the reason why 2-3% of patients in North America visit
their family doctor (80-83). In a recent study, patients with dyspepsia had
abnormal reflux as shown by 24hr pH monitoring (Small P.K., Gut; 1995; 36;
189). Very few endoscopic studies that describe the pathology of underlying
dyspepsia provide information about the frequency, or criteria for diagnosis
of esophageal pathology, particularly esophagitis (16). In this study, at
least 53% of patients with persistent dyspepsia, in spite of several weeks
of H2 blocker therapy, who had no reflux related symptoms and negative endoscopy,
were found to have RL. Presence of reflux in this group of patients was further
proved by biopsy evidence of chronic reflux esophagitis and favorable response
to more intensive antireflux therapy.
endoscopic finding worthy of consideration is the status of lower esophageal
sphincter (LES). A competent LES remains closed except when swallowing. During
endoscopy, without overinflating the esophagus with air, the fundus of the
stomach is seen in antegrade fashion when in midesophagus; then the presence
of incompetent or patulous (chalatic) sphincter is suspected. In 23 patients,
in addition to RL patulous LES was the only other endoscopic finding, 83%
of whom had GERD symptoms. Manometry, when performed in the majority of these
patients, showed LES pressure between 0-5mm of mercury (normal 10-25mm of
mercury). However, not every patient with manometric LESP less than 5mm of
mercury had endoscopic evidence or LESC. The prevalence of H&R and RL
was highest in patients with patulous LES and indeed 82% or these patients
had positive biopsy for chronic esophagitis. Therefore, the presence of LESC
during endoscopy is a significant positive finding indicating reflux.
is a common condition. H&R are common symptoms and when they are present
no further diagnostic testing is necessary before initiating empirical therapy
for symptomatic relief. However, less than 50% of patients with GERD complain
about heartburn. Endoscopy is indicated at least once in patients over the
age of 40 with persistent symptoms of reflux, in spite of adequate medical
treatment in order to look for Barrett's mucosa and to determine the extent
of esophageal mucal injury (grade I-IV modified Savary-Miller classifications).
This grading is more than theoretical and must be adopted by all endoscopists
since in addition to providing a blue print for comparison when further endoscopic
follow up is indicated and for evaluating new therapeutic modalities, it
also has important therapeutic implications (96A-1, A-2). Upper GI endoscopy
is negative for evidence of erosive in more than 50% of patients exhibiting
GERD related chest pain, ENT and pulmonary symptoms as well as H&R. Attention
to posterior larynx for evidence of erythema and edema, careful observation
of lower esophagus for presence of LESC and HH enhances diagnostic capabilities
of EGD and saves cost considerably.
in doubt, proper esophageal mucosa biopsy (A95) and proper histological study
of endoscopic biopsies confirms the diagnosis of up to 82% of GERD patients.
However, due to its expense and considerable degree of false positive and
false negative reporting, if morphometric studies are not done, biopsy should
be done when diagnosis is in doubt and in questionable cases.
studies suggest that proton pump inhibitors are the preferred form of medical
therapy. In addition to symptomatic relief of heartburn, they heal esophageal
mucosal injury and reduce the need for frequent dilatation in patients with
peptic stricture (9). Unfortunately, if treatment is stopped there is an
80% recurrence within six months (1). Since the basic mechanism of injury
is reflux of acid gastric contents in to the esophagus via an incompetent
lower esophageal sphincter (LES) (12), continuous long-term medical treatment
or surgical repair of the sphincter mechanism is recommended in severe cases