The incidence of esophageal adenocarcinoma is rising in the ENDOSCOPY CORNER

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2010;8:843 847 ENDOSCOPY CORNER Acetic Acid Spray Is an Effective Tool for the Endoscopic Detection of Neoplasia in Patients With Barrett s Esophagus GAIUS LONGCROFT WHEATON,* MOSES DUKU,* ROBERT MEAD,* DAVID POLLER, and PRADEEP BHANDARI* *Department of Gastroenterology and Department of Histopathology, Queen Alexandra Hospital, Portsmouth, United Kingdom See related article, Curvers WL et al, on page 1106 in Gastroenterology. BACKGROUND & AIMS: Diagnosis of Barrett s neoplasia requires collection of large numbers of random biopsy samples; the process is time consuming and can miss early-stage cancers. We evaluated the role of acetic acid chromoendoscopy in identifying Barrett s neoplasia. METHODS: Data were collected from patients with Barrett s esophagus examined at a tertiary referral center, between July 2005 and November 2008 using Fujinon gastroscopes and EPX 4400 processor (n 190). All procedures were performed by a single experienced endoscopist. Patients were examined with white light gastroscopy and visible abnormalities were identified. Acetic acid (2.5%) dye spray was used to identify potentially neoplastic areas and biopsy samples were collected from these, followed by quadrantic biopsies at 2 cm intervals of the remaining Barrett s mucosa. The chromoendoscopic diagnosis was compared with the ultimate histological diagnosis to evaluate the sensitivity of acetic acid chromoendoscopy. RESULTS: Acetic acid chromoendoscopy had a sensitivity of 95.5% and specificity of 80% for the detection of neoplasia. There was a correlation between lesions predicted to be neoplasias by acetic acid and those diagnosed by histological analysis (r 0.98). There was a significant improvement in the detection of neoplasia using acetic acid compared with white light endoscopy (P.001). CONCLUSIONS: Analysis of this large series showed that acetic acid-assisted evaluation of Barrett s esophagus detects neoplasia better than white light endoscopy, with sensitivity and specificity equal to that of histological analysis. Keywords: Chromoendoscopy; Dysplasia; Intramucosal Cancer; Endoscopic Mucosal Resection. View this article s video abstract at www.cghjournal.org. The incidence of esophageal adenocarcinoma is rising in the western world. It is responsible for 14,500 deaths per year in the United States 1 and 7000 per year in the United Kingdom. 2 Barrett s dysplasia is a well established precursor, which is the basis for endoscopic surveillance of patients with Barrett s esophagus. Standard protocols require the collection of large numbers of biopsies which are conventionally taken in a random quadrantic fashion at 2 cm intervals. Nontargeted multiple biopsies have remained standard practice as most of the dysplastic areas are difficult to see with standard resolution white light endoscopes, and only 13% of early neoplastic lesions appear as macroscopically visible nodules. 3 It has been shown that intensive surveillance biopsies as per the Seattle protocol using jumbo forceps does not improve the detection of intramucosal adenocarcinoma any more reliably than less intensive protocols with standard biopsy forceps. 4 Endoscopic surveillance of Barrett s esophagus makes clinical sense due to its malignant potential, but the poor pick up rate of neoplasia during routine surveillance with white light endoscopy (WLI) and quadrantic biopsy questions the cost effectiveness of this strategy. 5 Recent developments have led to improvements in the optical resolution of endoscopes, and new technologies have been developed which enhance mucosal visualization. Improved methods for identifying in vivo dysplastic areas use either chromoendoscopy, 6 8 of which acetic acid (AA) is 1 method, or narrow band imaging, 9 and spectral imaging techniques like Fujinon intelligent color enhancement. 10 Chromoendoscopy for the detection of Barrett s metaplasia has been available for many years. The 3 dyes commonly used are methylene blue, 11 indigo carmine, 12 and AA. 13 AA dye spray is used routinely for the detection of cervical precancerous lesions, including glandular lesions, at colposcopy. 14 AA was originally used in the esophagus as an aid to detect small segments of residual Barrett s metaplasia after ablation therapy. AA when sprayed on Barrett s mucosa leads to reversible acetylation of nuclear proteins leading to an aceto-white area, causing vascular congestion and improving the visualization of the mucosal surface. 15 This allows the mucosal surface patterns to be assessed, improving the diagnosis of dysplasia or cancer. The reaction only lasts a few minutes, with dysplastic tissue losing the acetowhitening more quickly than background Barrett s epithelium, further highlighting abnormal areas (see Figures 1 and 2). Our contention is that the detection of neoplasia in Barrett s is the main priority and not intestinal metaplasia. The aim of this study is to assess the accuracy of AA dye spray in the detection of Barrett s neoplasia during gastroscopy. Methods This study was approved by a central institutional review board, which allowed a waiver for patient-informed con- Abbreviations used in this paper: AA, acetic acid; EMR, endoscopic mucosal resection; HGD, high grade dysplasia; LGD, low grade dysplasia; WLI, white light endoscopy. 2010 by the AGA Institute 1542-3565/$36.00 doi:10.1016/j.cgh.2010.06.016

844 LONGCROFT WHEATON ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 10 Figure 1. HGD in Barrett s pre- (A) and post (B) AA dye spray. sent for data collection. Each patient had signed an informed consent form for the endoscopy procedure. Our department is a specialist center for the assessment and treatment of early gastrointestinal neoplasia, taking referrals from across the South of England for the endoscopic management of dysplasia and intramucosal cancer within Barrett s esophagus. All patients undergoing AA dye spray for evaluation of Barrett s esophagus between 2005 and 2008 were recorded prospectively on a computer database. All procedures were performed by a single experienced endoscopist (PB) with expertise in lesion recognition and endoscopic mucosal resection (EMR). The patient population included Barrett s cases where dysplasia had never been identified (group A) and tertiary referrals for suspected esophageal neoplasia where dysplasia had been found or suspected on random biopsy and post EMR follow-up cases where dysplasia had been treated endoscopically (group B). All patients were endoscoped on a dedicated chromoendoscopy list. Twenty-five percent of the cases were performed under local anesthetic with xylocaine 10 mg spray, with 75% choosing conscious sedation using intravenous midazolam, starting at 2.5 mg and increasing in 1 mg intervals as required to a maximum of 5 mg. Each procedure was allocated 15 20 minutes. Patients were excluded if they had severe esophagitis, contact bleeding, or acute mucosal trauma. Gastroscopy was performed using Fujinon EG-590zw and EG-590wr gastroscopes with the EPX 4400 processor (Fujinon, Tokyo, Japan). Magnification, if available, was not used for the detection of neoplasia. Patients Figure 2. A small focus of HGD seen after AA dye spray (circle). Figure 3. Pathway for AA chromoendoscopy. HRE, high-resolution endoscopy. were given 50 ml of a solution containing 5 ml of 10% N-acetyl cysteine and 5 ml of simethicone to drink prior to undertaking the procedure to act as a mucolytic and bubble-bursting agent. Endoscopy was then performed (Figure 3). Patients were first examined with conventional WLI, with any debris thoroughly removed and any visible abnormality suspicious of neoplasia noted (limb A of Figure 3). The length of the Barrett s segment was measured from the top of the gastric fold to the squamocolumnar junction, with islands of Barrett s recorded separately. Twenty ml of AA (2.5%) dye spray was then applied using a spray catheter to identify additional dysplastic foci within the Barrett s segment over and above that which was seen by WLI. (limb B of Figure 3). To help identify dysplasia after AA spray, an assessment of the following features was made: Surface pattern (ridged, villous, nodular, round, irregular); Mucosal vascular pattern (regular or irregular); and Acetowhitening reaction (normal or abnormal). Barrett s mucosa was classified by the endoscopist as nondysplastic if the surface pattern was round, tubular, villous, or ridged, the vascular pattern was normal, and the acetowhitening reaction was normal. 16 Mucosa was defined as dysplastic if the following was observed; irregular surface patterns, increased vascularity, an irregular microvascular pattern or early disappearance of acetowhitening from a focal area of mucosa as compared with the rest of the Barrett s mucosa. 16

October 2010 ACETIC ACID IN DETECTION OF NEOPLASIA IN BE 845 Mucosa was defined as invasive cancer if there was a complete loss of surface patterns, increased vascularity, disorganized and dense microvascular pattern, and rapid disappearance of the acetowhitening. In addition depressed areas (Paris type IIc) were regarded as suspicious for malignancy. After AA chromoendoscopy all lesions that became visible were noted and a targeted biopsy was taken and sent in a separate pot to pathology (limb C of Figure 3). This was followed by quadrantic biopsies at every 2 cm of the remaining Barrett s segment. These were sent separately for pathological review (nontargeted histology). If a suspicious lesion was seen on WLI, but appeared normal after AA dye spray it was biopsied as a part of the quadrantic biopsy strategy. In cases where an abnormality was seen on AA targeted biopsy, 4 quadrantic biopsies were not taken from the same quadrant at the same latitude as the targeted biopsy. The final histological diagnosis was achieved by combining the results of the targeted biopsies (limb C) and nontargeted quadrantic biopsies (limb D of Figure 3). All histological slides were reviewed by 2 expert pathologists. Analysis was performed on a per procedure basis, and not on a per lesion basis. Statistics were calculated using Microsoft Excel 2007 and MP associates MathCAD 2001i professional statistical package. Spearman s rank correlation coefficient (r) was calculated to assess the correlation between predicted and true histology. The proximity of r to 1 shows the strength of correlation, with r 1 being a 100% correlation. 2 test was used to assess the statistical significance of the difference between 2 limbs (WLI vs AA). P.05 was considered to be statistically significant. Results We performed 190 procedures on 119 patients with Barrett s metaplasia and neoplasia. Seventy-eight procedures were performed in group A (patients with no prior history of dysplasia) and 112 procedures were performed in group B (tertiary referral group). The median age of the patient cohort was 65 (range: 35 87) with 75% male. The median length of Barrett s segment was 4 cm (range: 2 15). True Neoplasia Histologically confirmed Barrett s neoplasia was found in 46% of the study population (88 out of 190 procedures). There were 21/88 cases of early cancer (T1a and T1b), 51/88 cases of high grade dysplasia (HGD), and 16/88 cases of low grade dysplasia (LGD). Table 2. Correlation Between Targeted Histology and Actual Histology Visible Abnormalities Visible neoplasia was noted by the endoscopist during 43/190 procedures with WLI endoscopy, and during 102/190 following AA dye spray. Use of WLI endoscopy alone, significantly under-diagnosed Barrett s neoplasia. AA dye spray significantly (P.001) improved the Barrett s neoplasia detection rate (2.5-fold) compared with WLI alone as shown in Table 1. Targeted Histology Histology results obtained from AA dye spray targeted biopsies were compared with the final pathological diagnosis (a combination of random and targeted biopsy results). An excellent correlation was identified (r 0.9; see Table 2). AA targeted biopsies diagnosed 63 of the 67 dysplasia cases. All 21 of the 21 cancers were diagnosed by AA targeted biopsies compared with 13 of the 21 with WLI. In some patients more than 1 area of dysplasia was found by targeted biopsy. However, in all cases where dysplasia was found on targeted biopsy there were no additional abnormalities found on random biopsy. Sensitivity and Specificity of AA Targeted Histology Four out of 88 Barrett s neoplasia cases were not identified on targeted histology, giving a false negative rate of 4.5% for all neoplasia. If the LGD population is excluded then the false negative rate falls to 2.7% (2/72). Twenty of 102 nondysplastic Barrett s mucosa, suspected to be neoplastic following AA dye spray because of visible abnormalities, turned out to be inflammation, giving a false positive rate of 19.6%. The overall sensitivity for identification of neoplasia was 95.5% with a specificity of 81%; see Table 3. On subgroup analysis we found a sensitivity of 77% and specificity of 85% for group A and a sensitivity of 98.7% and specificity of 75% for group B. Missed Neoplasia Targeted histology; group C Actual histology (random targeted); group D Dysplasia (HGD LGD) 63 67 Cancer 21 21 Correlation r 0.99 There were no missed cancers in either group. There were 4 missed cases of dysplasia; 2 LGD and 2 HGD. Three out of the 4 missed dysplasia were in group A. Two of these cases were classified as LGD and 1 was classified as HGD on the initial multiple quadrantic biopsies but were not identified on AA spray. All these patients have had at least 2 further Table 1. Visible Abnormalities: WLI Versus AA Spray Conventional WLI visible abnormality, limb A AA predicted in vivo histology, limb B B-A (P.001) Normal 147 88 59 (1.6-fold) Neoplasia 43 102 59 (2.3-fold) Table 3. Sensitivity, Specificity, and Prevalence in the Cohort Sensitivity (%) Specificity (%) Prevalence of neoplasia (%) Whole cohort 95.6 81 46.3 Group A 77 85 16.7 Group B 98.7 73 66.9

846 LONGCROFT WHEATON ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 10 gastroscopies at 6-month intervals with multiple further biopsies and no further evidence of dysplasia has been found. One patient had missed HGD in group B. This patient had received 2 previous EMRs and argon plasma coagulation ablation, and so had multiple islands of neosquamous epithelium. Discussion This is the largest series of AA chromoendoscopy reported, with the largest number of neoplasia cases being detected by this technique. This series shows that AA dye spray significantly improves the detection of neoplasia in Barrett s esophagus. It illustrates that AA targeted biopsies can diagnose dysplasia in the majority of patients, without the need of further multiple nontargeted biopsies. It also illustrates that no cancers are missed by AA targeted biopsies. Only 4 out of 88 patients with neoplasia were missed by AA dye spray; 2 of these were LGD, the other 2 were HGD. These mucosal dysplasias were picked up on conventional quadrantic biopsy. We feel that all of these cases were challenging both for the endoscopist and the pathologist. LGD is a difficult histological diagnosis to make, with an element of subjectivity involved, and at times inflammation can raise a possibility of LGD. We have a policy of repeating endoscopy in patients with LGD after 8 weeks of high dose proton pump inhibitor therapy (omeprazole 40 mg once daily). Both the patients have had 2 further gastroscopies on high dose acid suppression with multiple biopsies but no further LGD has been discovered, suggesting that the initial finding could be inflammation-related. Another patient from group A had HGD diagnosed on 1 of the several biopsies taken as a part of the quadrantic biopsy strategy. He has had 3 further gastroscopies and no further dysplasia has been found. This raises the possibility of microscopic monofocal HGD. The fourth patient with missed dysplasia was from group B and was found to have HGD. This was a challenging case with a history of 2 previous EMRs and argon plasma coagulation ablation resulting in multiple islands of neosquamous epithelium. The remaining 84 cases of Barrett s neoplasia were detected by targeted biopsy. Additional quadrantic biopsies did not alter the overall diagnosis. We believe that this is an important finding and it questions the logic of further quadrantic biopsies in patient groups where neoplasia is already detected by AA dye spray. This finding potentially has significant cost and resource implications for the endoscopist and the pathologist. Our findings are similar to another study quoting a surprisingly high sensitivity of 100% and specificity of 97.7%. 17 However, only 13/100 of the patients in the study had dysplasia, making it difficult to draw any meaningful conclusions from it with respect to dysplasia detection. While patients with excessive inflammation were excluded, a significant factor in the false positive rate of 19.6% was the presence of inflammatory changes appearing like dysplasia. We feel this highlights the importance of treatment with proton pump inhibitors prior to endoscopy. When Barrett s epithelium looks inflamed it is sometimes necessary to arrange further assessment after 8 weeks of a higher dosage of acid suppression, and this intervention should not be overlooked. There was a high prevalence of neoplasia in group A of 16.7%. In a Barrett s surveillance population this would be expected to be much lower. However, it should be noted that this population was not a standard surveillance population. It consisted of Barrett s surveillance cases and symptomatic patients who had no prior diagnosis of dysplasia who had been listed for assessment on a chromoendoscopy list. Therefore it is not surprising that the prevalence of dysplasia in this cohort is higher than a standard asymptomatic surveillance population. Group B is a complex group, consisting of all patients with previously diagnosed dysplasia being assessed for EMR and post EMR follow-up patients to look for metanchronous neoplasia. Endoscopic treatment in this cohort depends on accurate localization of neoplasia, and we were reassured to find that dysplasia could be targeted with a high degree of accuracy in this cohort. Electronic imaging techniques like narrow band imaging, and spectral imaging (Fujinon intelligent color enhancement) have been reported in the evaluation of Barrett s neoplasia. 18,19 However, the studies have been very small and it is not possible to draw any firm conclusions from these studies. A problem with all virtual chromoendoscopy techniques is that they have cost and resource implications as they require commitment to a particular endoscope manufacturer, and need the most recent image processors and endoscopes. Many units (even in the Western world) do not have the latest models of processor and endoscopes, and therefore cannot use this technology. AA dye spray can be performed in any unit and is compatible with all makes and models of endoscope, irrespective of the manufacturer. This is a single-center study, where all endoscopies have been performed by a single experienced endoscopist (P.B.). The results are therefore not generalizable to all endoscopists. It is probable that to achieve similar results formalized training would need to be undertaken in lesion recognition. This is an accepted part of endoscopy, and we believe that if dye spray is to be used widely in the assessment of Barrett s it will be necessary to address this issue. However, we believe that intensive training in a large volume expert center can help achieve the expertise in a relatively short period of time. Some of the patients in group B have had more than 1 procedure performed on them. This is due to our policy to follow up patients post EMR with 3 monthly gastroscopies to look for metanchronous neoplasia. Therefore procedures were evaluated rather than patients. This could be considered a potential limitation but we believe otherwise as these are the most challenging patients to identify dysplastic foci due to the presence of multiple neosquamous islands and distorted anatomy. Patients do develop dysplasia between endoscopies, which is why they are followed up. Therefore previous endoscopy results should not influence the findings of the next procedure. It is important that if we perform a targeted treatment we can safely follow this cohort up, which is reflected in this study. Our study is not a randomized controlled trial, and as such we cannot say how much of the dysplasia would have been found or missed by random biopsies had dye spray not been used. However, our data does show that dysplasia was visible in only 50% of our patients on high resolution WLI. Could protocol-guided quadrantic biopsies every 2 cm be able to pick up the dysplasia in the remaining 50% of patients with invisible (WLI) dysplasia? This is a difficult question to answer from the design of our study but we believe that it would be very difficult to diagnose the dysplasias by quadrantic biopsies alone in this

October 2010 ACETIC ACID IN DETECTION OF NEOPLASIA IN BE 847 group of patients with invisible dysplasia. Even if it does, then it would require a very large number of biopsies with significant resource implications. Alternatively, the endoscopist could elect to use AA spray and that will help better define the dysplasia that is already seen on WLI and also highlight the invisible (WLI) dysplasia as seen in 50% of the dysplastic population. Our data are not enough to completely abandon the strategy of protocol-guided biopsies in all patients with Barrett s but it does justify the role of AA in patients with invisible dysplasia on WLI and also questions the additional gain of multiple nontargeted biopsies after dysplasia has been seen with AA spray. Conclusions Our study demonstrates that AA targeted biopsies are an excellent tool to localize prevalent neoplasia within Barrett s esophagus. It is cheap, quick, universally available, and effective. It questions the relevance of additional nontargeted biopsy in patients where AA has already identified neoplasia. Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at www.cghjournal.org, and at doi:10.1016/ j.cgh.2010.06.016. References 1. Jemal A, Siegel R, Ward E, et al. Cancer statistics. CA Cancer J Clin 2009;59:225 249. 2. Barr H. The case against oesophageal resection. Ann R Coll Surg Engl 2007;89:586 588. 3. Pech O, Gossner L, Manner H, et al. Prospective evaluation of the macroscopic types and location of early Barrett s neoplasia in 380 lesions. Endoscopy 2007;39:588 593. 4. Kariv R, Plesec TP, Goldblum JR, et al. The Seattle protocol does not more reliably predict the detection of cancer at the time of esophagectomy than a less-invasive surveillance protocol. Clin Gastroenterol Hepatol 2009;7:653 658. 5. Macdonald CE, Wicks AC, Playford RJ. Final results from 10 year cohort of patients undergoing surveillance for Barrett s oesophagus: observational study. BMJ 2000;321:1252 1255. 6. Sharma P, Weston AP, Topalovski M, et al. Magnification chromoendoscopy for the detection of intestinal metaplasia and dysplasia in Barrett s oesophagus. Gut 2003;52:24 27. 7. Gossner L, May A, Pech O, et al. Chromoendoscopy for detection of dysplasia or mucosal cancer in Barrett s esophagus. Gastrointestinal Endoscopy 2000;51:A3537. 8. Fortun PJ, Anagnostopoulod GK, Kaye P, et al. Acetic acid-enhanced magnification endoscopy in the diagnosis of specialized intestinal metaplasia, dysplasia and early cancer in Barrett s oesophagus. Aliment Pharmacol Ther 2006;23:735 742. 9. Wolfsen HC, Crook JE, Krishna M, et al. Prospective, controlled tandem endoscopy study of narrow band imaging for dysplasia detection in Barrett s esophagus. Gastroenterology 2008;135: 24 31. 10. Miwako A, Masahiro T, Hideaki A. Magnifying endoscopy in the diagnosis of Barrett s esophagus and Barrett s adenocarcinoma. Esophagus 2007;3:165 169. 11. Canto M, Selrakian S, Willis J, et al. Methylene-blue directed biopsies improves detection of intestinal metaplasia and dysplasia in Barrett s oesophagus. Gastrointestinal Endoscopy 2000; 51:560 568. 12. Szaloki T. Indigo carmine contrast staining in combination with high resolution electronic endoscopy. Gasztroenterológiai Osztály 2002;143:25 29. 13. Hoffman A, Kiesslich R, Bender A, et al. Acetic acid-guided biopsies after magnifying endoscopy compared with random biopsies in the detection of Barrett s esophagus: a prospective randomized trial with crossover design. Gastrointestinal Endoscopy 2006;64:13 16. 14. Van Le L, Broekhuizen FF, Janzer-Steele R, et al. Acetic acid visualization of the cervix to detect cervical dysplasia. Obstet Gynecol 1993;81:293 295. 15. Lambert R, Rey JF, Sankaranarayanan R. Magnification and chromoscopy with the acetic acid test. Endoscopy 2003;35:437 445. 16. Réaud S, Croue A, Boyer J. Diagnostic accuracy of magnifying chromoendoscopy with detection of intestinal metaplasia and dysplasia using acetic acid in Barrett s esophagus. Gastroenterol Clin Biol 2006;30:217 223. 17. Vázquez-Iglesias JL, Alonso-Aguirre P, Diz-Lois MT, et al. Acetic acid allows effective selection of areas for obtaining biopsy samples in Barrett s esophagus. Eur J Gastroenterol Hepatol 2007; 19:187 193. 18. Sharma P, Bansal A, Marthur S, et al. The utility of a novel narrow band imaging endoscopy system in patients with Barrett s oesophagus. Gastrointestinal Endoscopy 2006;64:167 175. 19. Pohl J, May A, Rabenstein T, et al. Comparison of computed virtual chromoendoscopy and conventional chromoendoscopy with acetic acid for detection of neoplasia in Barrett s esophagus. Endoscopy 2007;39:594 598. Reprint requests Address requests for reprints to: Dr Pradeep Bhandari, Department of Gastroenterology, Queen Alexandra Hospital, Cosham, Portsmouth PO6 3LY, United Kingdom. e-mail: deep3570@yahoo.co.uk; fax: (44) 7985442226. Conflicts of interest The authors disclose no conflicts.