Oral Epithelial Dysplasia: Causes, Diagnosis, and Treatment

Oral epithelial dysplasia is a microscopic finding in which the cells lining the inside of the mouth show abnormal growth patterns that can, in some cases, progress to oral squamous cell carcinoma. It is the most important histological marker pathologists use to assess whether a white patch, red patch, or other suspicious oral lesion carries cancer risk. Roughly one in five patients with dysplastic oral lesions will develop cancer within five years, though that number varies considerably depending on the grade of dysplasia, where the lesion sits, and who the patient is.

What Causes Oral Epithelial Dysplasia

Tobacco and alcohol are the two most thoroughly established risk factors. Smoking, particularly unfiltered cigarettes, independently raises the likelihood of developing dysplastic changes in the oral lining. Alcohol consumption does too, and the two habits compound one another: the combined effect is greater than either habit alone, though exclusive tobacco use carries more risk than exclusive alcohol use.¹ This synergy matters because it means a person who both smokes and drinks heavily is at substantially higher risk than someone who does just one or the other.

In South and Southeast Asia, betel quid chewing is a major driver. The areca nut in betel quid releases reactive oxygen species in the chewer’s saliva during natural chemical breakdown, directly damaging oral cells. Areca-nut-specific nitrosamines formed during this process are mutagenic and capable of inducing tumors in animal experiments.² Chronic exposure to areca nut compounds also triggers DNA damage, inflammation, tissue oxygen depletion, and changes that push normal cells toward a more invasive behavior.³

Human papillomavirus, particularly the high-risk strains also linked to cervical cancer, represents a distinct pathway. HPV-associated oral dysplasia has its own characteristic microscopic appearance, including cells with halos around shrunken nuclei (a sign of active viral replication) and other features tied to viral activity. These lesions have been described under different names over the years, including oral koilocytic dysplasia and HPV-associated intraepithelial neoplasia, and researchers are still working out whether they should be classified as a separate entity from conventional oral dysplasia.

How Oral Dysplasia Is Detected

You cannot diagnose oral epithelial dysplasia just by looking at the mouth. A white patch (leukoplakia) or red patch (erythroplakia) may look suspicious, but the only way to confirm dysplasia is by taking a tissue sample and examining it under a microscope. That biopsy is the gold standard and the step everything else revolves around.

Several adjunctive screening tools exist to help clinicians decide which lesions need a biopsy, though none replace it. Autofluorescence devices shine a specific wavelength of light into the mouth; abnormal tissue fluoresces differently from healthy tissue. In studies, autofluorescence picked up dysplastic changes with good sensitivity (around 84%) but poor specificity (around 15%), meaning it catches most cases but also flags many lesions that turn out to be benign.⁴ That trade-off is acceptable as a screening tool (you would rather over-biopsy than miss a cancer), but it means many patients will get biopsies they did not need.

Toluidine blue staining, where a dye is painted onto the oral mucosa and selectively retained by abnormal cells, offers a different balance. One study found a sensitivity of 95% and specificity of about 71%, with overall diagnostic accuracy of roughly 86%.⁵ However, performance varies across studies. A more recent cross-sectional study found toluidine blue sensitivity closer to 58% and specificity of 85%.⁶ The gap likely reflects differences in the populations studied and the types of lesions included. In practice, clinicians sometimes combine methods, though the added accuracy of combining tools remains modest.

Grading Systems and Why They Matter

Once a biopsy confirms dysplasia, the pathologist grades it, because the grade shapes every decision that follows. The World Health Organization’s system, most recently updated in 2017, divides dysplasia into three tiers: mild, moderate, and severe (with carcinoma in situ at the extreme end). A simpler binary system groups cases as either low-risk or high-risk. Both systems are in wide use, and one longstanding question is whether either does a better job predicting which lesions will become cancer.

The answer, based on the available evidence, is that they perform similarly. A systematic review and meta-analysis found no significant difference between the WHO three-tier system and the binary system in predicting malignant transformation.⁷ A separate study looking specifically at squamous cell carcinoma development confirmed that the two grading approaches had similar predictive value.⁸

Where the binary system does pull ahead is consistency. Pathologists looking at the same slide often disagree on the grade, and that disagreement is worse with the three-tier WHO system. A meta-analysis pooling observations from 46 pathologists across eight studies found that agreement between different pathologists was roughly twice as high for the binary system as for the WHO system. Agreement when the same pathologist re-examined their own slides was also better with the binary approach.⁹ The practical implication is that the binary system reduces the chance that a patient’s management changes simply because a different pathologist looked at the biopsy. That said, the WHO system remains the standard framework in most clinical guidelines.

How Often Dysplasia Becomes Cancer

This is the question patients care about most, and the honest answer is: it depends heavily on the individual case. The overall malignant transformation rate across all grades is commonly estimated at somewhere between 5% and 20%, with follow-up periods varying across studies. One study estimated that about 22% of patients with oral dysplasia underwent malignant transformation within five years.¹⁰ In a cohort of 46 patients followed for three years, about 8% developed squamous cell carcinoma confirmed by second biopsy.¹¹

Grade matters, but not as cleanly as you might expect. High-grade dysplasia is an independent risk factor for transformation to cancer.¹² One meta-analysis found that lesions classified as severe dysplasia or carcinoma in situ by the WHO system had a pooled malignant transformation rate of about 40%, while high-risk lesions under the binary system transformed at a pooled rate of about 31%.⁷ But even mild dysplasia can progress, and severe dysplasia sometimes stays put for years. That unpredictability is part of why the condition is so difficult to manage.

Other features that predict progression include the location of the lesion, its size (larger than roughly 200 mm²), and whether the lesion has a non-homogeneous appearance, such as mixed red and white patches or an irregular surface. Counterintuitively, non-smokers with dysplasia had a higher transformation rate in at least one large study, possibly because dysplasia in a non-smoker may signal a more inherently aggressive biology rather than a reversible toxic injury.¹⁰

Treatment Options

Treatment broadly falls into three strategies: surgical removal, non-surgical interventions, and watchful waiting with regular monitoring. None of these has been shown in large randomized trials to definitively prevent cancer, which is a gap in the evidence that frustrates clinicians and patients alike.

Surgery, whether by conventional scalpel or laser, is the most common active treatment for moderate-to-severe dysplasia. Pooled data suggest that the malignant transformation rate after scalpel surgery is about 9%, after laser therapy about 6%, and with clinical observation alone about 10%.¹³ The difference between surgery and observation is real but modest. In a UK study focused on high-grade dysplasia, about 29% of untreated patients developed cancer compared to about 12% of those who had their lesions excised.¹⁴ That halving of risk is meaningful, but the fact that one in eight treated patients still developed cancer underscores surgery’s limits.

Photodynamic therapy is a less invasive option in which a light-sensitive drug is applied to the lesion and then activated with a specific wavelength of light, destroying the abnormal cells. Early results suggest it works best for low-grade dysplasia and carcinoma in situ, and patients generally tolerate it well.¹⁵ It is minimally invasive, requires little preparation, and has high patient acceptance, making it a promising option especially where surgery would be difficult or disfiguring.¹⁶ However, larger trials are needed before it becomes standard practice.

For mild dysplasia, many clinicians opt for close monitoring rather than immediate intervention, scheduling regular follow-up exams and rebiopsy if the lesion changes. Risk factor modification, particularly quitting tobacco and reducing alcohol intake, is part of every management plan regardless of grade.

Why Lesions Come Back After Treatment

One of the more frustrating aspects of oral dysplasia is its tendency to recur even after apparently complete surgical removal. This is partly explained by a concept called field cancerization. The idea is that the visible lesion is just the tip of the iceberg: a broader area of the oral lining has already accumulated genetic damage, even though it looks normal to the naked eye. When a surgeon removes the obvious lesion, those genetically altered cells in the surrounding tissue can independently give rise to new abnormal patches or even cancer.¹⁷

Field cancerization also helps explain why some patients develop multiple separate oral cancers over time. Oral squamous cell carcinoma has a high rate of local failure and second primary tumors, and the existence of these genetically altered fields surrounding the original tumor is considered a key reason.¹⁸ This is why lifelong surveillance matters even after successful treatment: the problem is never just the single spot that was removed.

Diagnostic Challenges and Look-Alikes

Diagnosing oral dysplasia under the microscope is not always straightforward, even for experienced pathologists. One common difficulty is that dysplasia can trigger a dense lymphocytic immune response in the tissue, making it look very similar to a benign condition called lichenoid mucositis. When the immune response dominates the microscopic picture, the underlying dysplastic changes can be obscured.¹⁹ Getting this distinction wrong has real consequences: misdiagnosing dysplasia as a benign inflammatory condition delays appropriate management.

Another complication is reactive atypia. Chronic inflammation from things like ill-fitting dentures or persistent ulcers can cause oral cells to look abnormal under the microscope in ways that mimic true dysplasia. These reactive changes are a response to injury, not a step toward cancer, but telling them apart from genuine dysplasia can be difficult.²⁰ Some researchers have proposed that what is currently labeled “reactive” atypia might, in some cases, represent genuine early dysplasia that is being dismissed. The boundary between the two remains a genuine area of uncertainty in oral pathology.

The Immune Microenvironment

Recent work has begun to clarify what the immune system is doing in and around dysplastic lesions, and some of the findings have been surprising. One study examining gene expression in oral potentially malignant disorders found that the most significant changes happened not in the epithelial cells themselves but in the surrounding stromal tissue, and most of those changes related to immune activity. Macrophages, a type of immune cell, were the dominant infiltrating population, especially at the dysplasia stage. Unexpectedly, gene signatures associated with a specific macrophage subtype (often considered pro-tumor in other cancers) were independently linked to better oral cancer-free survival in patients with leukoplakia.²¹ This challenges the simplistic view that immune infiltration in a precancerous lesion is always bad news, and it opens the door to potentially using immune profiles to better predict which lesions are dangerous.

Molecular Markers on the Horizon

Because grading under the microscope is inherently subjective, researchers have long hoped that molecular markers could offer more objective risk assessment. The protein p53, sometimes called the “guardian of the genome” because of its role in detecting and responding to DNA damage, has been the most studied candidate. A high proportion of dysplastic lesions show abnormal p53 expression compared to non-dysplastic tissue, and expression in the upper cell layers of the oral lining appears to be a risk marker for eventual cancer development regardless of the histological grade assigned.²² However, no single molecular marker has yet proven reliable enough to replace conventional microscopic grading in routine clinical practice. The field is moving toward panels of multiple markers, but widespread clinical adoption remains a work in progress.

Who Gets Oral Dysplasia

Oral potentially malignant disorders as a group are most common in people aged 50 and older, with men making up about 63% of cases globally. Prevalence varies by region: Asia has the highest rates (around 5.4%), followed by South America and the Caribbean (around 5.2%) and Europe (around 4.9%).²³ The high Asian prevalence reflects the widespread use of betel quid in that region. When looking specifically at confirmed oral epithelial dysplasia (rather than the broader category of potentially malignant disorders), one large study found a mean patient age of 59 years and a male-to-female ratio of 0.67, meaning women actually outnumbered men in that dataset.²⁴ The demographic profile varies depending on the population and the risk factors prevalent in a given region.

Living with an Oral Dysplasia Diagnosis

The psychological burden of living with a condition that might become cancer is significant and often underrecognized. A study specifically examining psychosocial impacts of oral dysplasia found that about 30% of patients had anxiety symptoms, 16% had depressive symptoms, and 26% showed emotional distress. Nearly 70% experienced anxiety specifically related to the procedures they knew they would face as part of ongoing management, such as repeated biopsies with local anesthetic injections. More than 40% reported daily problems related to their oral health.²⁵

This is worth highlighting because follow-up for oral dysplasia is typically long-term, sometimes lifelong, and the anxiety does not go away after the first biopsy comes back stable. Experts recommend tailored follow-up schedules based on individual risk, with more frequent monitoring for high-grade dysplasia and other high-risk features.²⁶ The emphasis on patient-centered protocols reflects a growing awareness that managing the person, not just the lesion, matters for long-term outcomes and quality of life.

Artificial Intelligence in Dysplasia Grading

Given how much pathologist disagreement shapes dysplasia grading, AI-assisted analysis of biopsy slides is an active area of development. Multiple deep learning architectures have been tested on the task of detecting and grading oral epithelial dysplasia from digital images of tissue slides. One comparative study of nine architectures found that the best-performing model achieved about 95% accuracy, with particularly strong performance in distinguishing normal tissue from mild and severe dysplasia.²⁷ A systematic review of the field confirmed that several AI approaches have reached accuracies above 90% in experimental settings.²⁸

The practical appeal is obvious: AI does not have interobserver variability. It will grade the same slide the same way every time. Another study reported diagnostic accuracy above 93% across categories, with the strongest agreement in non-dysplastic and severe categories and moderate discrepancies in the mild-to-moderate range, which is exactly where human pathologists struggle most.²⁹ The consensus is that AI has strong potential as a screening and second-opinion tool, particularly in high-volume settings, though human oversight will remain essential for the foreseeable future. Most existing studies are experimental, tested on curated datasets rather than in the messy reality of daily diagnostic pathology, and the jump from lab performance to clinical deployment is not trivial.