Lack of Iodine Staining Lugol’s Chromoendoscopy Predicts Squamous Neoplastic Progression in a High-risk Region of China: Implications for East and West

Esophageal squamous cell carcinoma (ESCC) is the most common global esophageal malignancy and the sixth leading cause of cancer mortality, accounting for 90% of the 456,000 incident esophageal cancers each year.^1,2 An Asian Esophageal Cancer Belt stretching from Turkey to central China encompasses several discrete geographic areas with high incidence of esophageal cancer where age-adjusted mortality rates are greater than 40 per 100,000 per year, greater than the age-adjusted mortality rate of any cancer site in the United States.^3,4 Informed by observational studies in the region demonstrating a high local rate of disease, screening for ESCC has been attempted in high-risk regions of China for ESCC since 2005.^5,6

Lugol’s iodine chromoendoscopy (LCE) uses Lugol’s solution, a solution of iodine and potassium iodide which stains alpha-1,4 glucans in glycogen, turning normal squamous epithelium dark brown, while leaving squamous cell dysplasia and carcinoma unstained.⁷ LCE has been studied to target histologic sampling for diagnosis of ESCC and squamous dysplasia, acting as a negative stain, where absence of staining correlates with histologic detection of ESCC and severe squamous dysplasia on biopsy of the unstained area.^8,9 Studies comparing LCE to biopsy have estimated the sensitivity for squamous dysplasia and cancer from 46–100% and the specificity from 40–90%.^10–12 Two large studies have shown a high sensitivity of LCE compared to biopsy for severe squamous dysplasia and ESCC in high-risk regions of China.^13,14

In this issue of Clinical Gastroenterology and Hepatology, an insightful analysis within the intervention arm of an ongoing, large, cluster-randomized, controlled trial from Dr. Lui et al in the Hua county region of China (a high-risk region for ESCC), has contributed significantly to the prior literature in elucidating the testing characteristics of Lugol’s chromoendoscopy for ESCC and dysplasia in the Asian Esophageal Cancer belt. The authors found that participants with unstained areas on LCE but biopsies without squamous dysplasia had a higher rate of progression to severe squamous dysplasia, squamous carcinoma in situ, or invasive squamous cell carcinoma than those with mild or moderate dysplasia on non-targeted biopsy and the absence of unstained lesions on Lugol’s iodine staining. This finding suggests the need for a substantial change in screening practice in China. Currently, subjects with severe dysplasia or early stage cancer on screening exam are treated with endoscopic mucosal resection with or without chemotherapy and radiation, subjects with mild to moderate dysplasia would be re-examined in three years, and subjects with normal histopathology (even with unstained areas on LCE) would not have any further screening.¹⁵ However, if screening interventions for ESCC (i.e. LCE) do have a clinical benefit, for which a trial by the same authors is underway, it follows that subjects with unstained lesions but negative biopsies should have a follow-up endoscopy.¹⁶

The authors’ findings are objectively important, affecting a significant proportion of the patients in their chosen population. Among the 1,468 participants in the screening arm included for this analysis, 903 (62%) had unstained lesions but no histologic dysplasia, while 357 (24%) had mild or moderate dysplasia. Under the current guidelines, only these 357 would have had follow-up screening.¹⁶ Moreover, the group of 903 was just a small proportion (6%) of the overall patients screened, and all 1,468 with any positive finding were just under 10% of patients screened. This suggests that, while the group of subjects identified in this study as requiring follow-up is substantial, it would likely not be to an economically prohibitive level.

In response to this finding, the authors propose a more nuanced approach to risk-stratify patients with an intermediate result in their screening exam using a predictive modeling approach. The authors chose an appropriate subset of patients, those with an unstained lesion on LCE or mild to moderate dysplasia on biopsy not requiring treatment, in which to develop their predictive model. The patients with neither unstaining area nor any dysplasia on biopsies (i.e the lowest-risk group after screening) and subjects with high-grade dysplasia or squamous cell cancer (i.e. the highest-risk group after screening) were excluded. They also excluded subjects with dysplasia on non-targeted biopsy because only LCE-targeted biopsy is performed in local screening practice. They validated their predictive model with clinically incident ESCC among patients that did not accept a follow-up exam. By including histology findings, unstained lesion category, age, and BMI, the authors found an area under the receiver-operator characteristic curve of 0.868, and 0.850 in the validation subset. This suggests very good discrimination, with successful separation of high-risk subjects from low-risk subjects.

There could be several explanations for the observation that the presence of unstained lesions is more predictive of neoplastic progression than non-targeted positive biopsy results. Subjects without unstained lesions had biopsy of two standardized locations in the normal-appearing esophagus. Biopsy of normal appearing mucosa has a low pre-test probability of true neoplasia, and mild to moderate dysplasia may be nonspecific; thus, this combination would lead to a low post-test probability of true neoplasia despite a positive biopsy. By contrast subjects with unstained lesions have a high pre-test probability of a true neoplastic lesion, and non-targeted biopsy may be insensitive due to sampling error, leading to a high post-test probability of true neoplasia in this group despite a negative biopsy. In this group, what is interpreted as progression on the follow-up visit would actually be a delayed diagnosis. Another possible pathophysiologic explanation for the authors’ findings is that biopsies of unstained lesions have high sensitivity, and even if true dysplasia is absent, the unstained lesions still indicate risk of development of neoplasia in a way not captured by biopsy. In either case, follow-up endoscopy for patients with unstained lesions but negative histology is supported by the authors findings.

While this study is promising and clinically relevant to the residents of the Hua County, there are several important limitations. There were a small number of events, 28 in the development cohort and 11 in the validation cohort, which limited statistical precision. The authors determined the pattern of Lugol’s iodine staining from retrospective, blinded review of endoscopic images and, while they used endoscopic photographs from a standardized endoscopic position, examination during the endoscopy would likely be more accurate. The very high risk setting likely facilitated the study by making progression common enough to study with reasonable precision, but it likely limited generalizability of the findings to low risk settings. Further, even in high risk settings the optimal timing and duration of surveillance is unknown. Given the relatively low rate of progression in these patients, observational studies with large numbers and long-term follow-up will likely be required to determine the optimal surveillance timeline.

This study also raises questions about whether Lugol’s staining should be done more in high risk patients in the West. In the United States, screening is considered in patients with tylosis, achalasia, squamous cell carcinoma of the head and neck, and in patients with history of caustic injuries to the esophagus.^17,18 However, since the molecular subtypes of ESCC are geographically and etiologically heterogeneous, it is possible that the same findings of Hua County will not be applicable to Western patients.¹⁹ If generalizable, though, it might be prudent for Western endoscopists who screen patients in these settings to consider performing Lugol’s staining during upper endoscopy, with unstained lesions as an indication for follow-up endoscopy regardless of histopathology diagnosis.

In the future, further scientific investigation is warranted to better understand the mechanisms of development of squamous cell carcinoma of the esophagus. The loss of iodine staining in dysplasia suggests glycogen depletion occurs early in the development of ESCC; as such, a deeper understanding of the metabolic changes occurring in early tumorigenesis may both yield new biomarkers of disease as well as therapeutic targets.^20,21 Finally, the role of the microbiome is increasingly understood to contribute to the development ESCC,²² and its role in the alteration of the ESCC metabolome has yet to be determined.