📄 medicine

Diagnostic Accuracy of Colonoscopy and Histopathological Yield of Routine Biopsy: A Prospective Cohort Study in a Tertiary Care Hospital

This prospective cohort study demonstrates that colonoscopy exhibits substantial concordance with histopathology and excellent accuracy for malignancy detection, while also validating the clinical utility of routine biopsy in macroscopically normal mucosa, which revealed significant inflammatory pathology in approximately one-fifth of cases.

Original authors: VINAYAK KSHIRSAGAR, VINAY BADANGI, AMIT GIRME, GARVIT SHARMA, SIMRAN SHAH

Published 2026-07-01

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: VINAYAK KSHIRSAGAR, VINAY BADANGI, AMIT GIRME, GARVIT SHARMA, SIMRAN SHAH

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine your colon is a long, winding hallway inside your body. When something goes wrong with your digestion, doctors use a special camera on a flexible tube (a colonoscopy) to walk down that hallway and take a look. Usually, what the doctor sees with their eyes is enough to make a guess about what's wrong. But sometimes, the hallway looks perfectly clean and smooth, even though the patient is still feeling sick.

This study is like a "reality check" for those doctors. They asked: If the hallway looks clean, is it actually clean? And how often does the doctor's "eye test" match the "lab test"?

Here is the story of what they found, broken down into simple parts:

1. The Setup: The "Eye Test" vs. The "Microscope Test"

The researchers followed 323 people who went to a hospital because of stomach trouble. Every single person got a colonoscopy.

The Eye Test: The doctor looked at the colon and wrote down what they thought it was (e.g., "Looks like cancer," "Looks like a polyp," or "Looks normal").
The Microscope Test: The doctor took tiny samples (biopsies) from the colon walls and sent them to a lab. The lab looked at the cells under a microscope to give the real diagnosis.

The study compared these two reports to see how often they agreed.

2. The Big Reveal: The "Clean" Hallway Isn't Always Clean

One of the most interesting findings was about the people whose colonoscopy looked perfectly normal.

The Surprise: Even though the camera saw nothing wrong in 140 patients, the microscope found something wrong in about 1 out of every 5 of them (21.4%).
The Analogy: Imagine walking into a room that looks spotless. You think, "Everything is fine." But if you put on special glasses (the microscope), you might see a layer of dust or a tiny stain you couldn't see before. In this study, that "dust" was mostly mild inflammation (colitis).
The Takeaway: Just because the colon looks normal doesn't mean it's healthy. Taking those tiny samples is still a good idea, even when the camera sees nothing.

3. How Good Was the Doctor's "Eye Test"?

The researchers wanted to know: Can the doctor trust their own eyes when they see something scary, like a tumor?

The Result: The doctor's eyes were extremely sharp when it came to spotting cancer.
- If the doctor said, "That looks like a tumor," they were right almost every time.
- If the doctor said, "That doesn't look like a tumor," they were 100% correct. There were zero "missed" cancers in this group.
The Catch: While the doctor was great at spotting cancer, they weren't perfect at guessing exactly what a non-cancerous lump was. Sometimes a benign (harmless) growth looked scary, and the lab had to step in to say, "Don't worry, it's just a harmless bump."

4. The "Red Flags" for Cancer

The study also tried to figure out who is most likely to have cancer before they even get the camera inside. They looked at two main clues:

Age: Being over 50 years old.
The Look: Seeing a visible "mass" or lump during the colonoscopy.

The Analogy: Think of it like a weather forecast.

If you see a giant dark storm cloud (a mass lesion) and it's late in the season (over 50 years old), the chance of a storm (cancer) is huge.
The study found that if you have a mass, your odds of having cancer are nearly 90 times higher than if you don't.
Interestingly, just having "rectal bleeding" (blood in the stool) wasn't a strong enough clue on its own to predict cancer without seeing the mass or knowing the age.

5. The Verdict

The Agreement: The doctor's visual guess and the lab's microscope report agreed with each other about 83% of the time. That's a very strong match.
The Routine: Even when the colon looks perfect, taking a tiny sample is worth it because it catches hidden inflammation in about 20% of people.
The Warning Signs: If a patient is over 50 and has a visible lump, that is a very strong signal that cancer might be present.

Summary in One Sentence

This study confirms that while a doctor's eyes are excellent at spotting cancer, they aren't perfect at seeing everything (like hidden inflammation), so taking tiny samples (biopsies) is still a smart move even when the colon looks perfectly clean.

Technical Summary: Diagnostic Accuracy of Colonoscopy and Histopathological Yield of Routine Biopsy

Problem Statement
Colonoscopy is the established gold standard for investigating lower gastrointestinal symptoms, yet the diagnostic value of performing routine biopsies on macroscopically normal mucosa remains a subject of debate. While endoscopic visualization guides immediate management, histopathology serves as the final arbiter of diagnosis. There is significant variability in reported concordance rates between colonoscopic impressions and histopathological findings. Furthermore, it is unclear whether routine sampling of unremarkable mucosa yields clinically relevant pathology (such as microscopic colitis or early inflammatory bowel disease) or if selective biopsy strategies are more appropriate. Additionally, identifying robust clinical and endoscopic predictors of malignancy is critical for prioritizing patients, particularly in resource-constrained settings.

Methodology
This was a prospective observational cohort study conducted at a tertiary care center in Pune, India, between January 2024 and March 2026. The study included 323 consecutive patients undergoing colonoscopy with biopsy for lower gastrointestinal symptoms.

Data Collection & Categorization: For each patient, the indication, colonoscopic impression, and final histopathological diagnosis were recorded. To facilitate structured comparison, both colonoscopic impressions and histopathological diagnoses were independently classified into six mutually exclusive categories: (1) Malignancy/growth, (2) TB/granulomatous, (3) Inflammatory bowel disease (IBD), (4) Colitis/inflammatory/non-specific ulceration, (5) Polyp, and (6) Normal/no significant pathology.
Statistical Analysis: Concordance between colonoscopy and histopathology was assessed using simple percentage agreement and Cohen's kappa ( $\kappa$ ). Diagnostic accuracy (sensitivity, specificity, PPV, NPV) of colonoscopic impression for malignancy was calculated against histopathology as the reference standard. Multivariable logistic regression, utilizing bootstrap resampling to account for the relatively small number of malignancy events ( $n=32$ ), was employed to identify independent predictors of malignancy. The discriminative performance of models was evaluated using the Area Under the Curve (AUC).

Key Results

Demographics: The cohort had a mean age of 43.4 years (range 8–81) with a marked male predominance (3.25:1).
Findings Distribution: Colonoscopy was abnormal in 56.7% of patients. Histopathology revealed no significant pathology in 34.4%, non-specific colitis/inflammatory changes in 30.0%, hyperplastic polyps in 11.1%, malignancy in 9.9%, IBD in 4.3%, and TB/granulomatous pathology in 2.2%.
Concordance: Overall agreement between colonoscopic impression and histopathology was 82.6% ( $\kappa = 0.761$ ), indicating substantial concordance. The primary source of discordance involved patients with a normal colonoscopic impression who were found to have histological colitis.
Diagnostic Accuracy for Malignancy: Colonoscopy demonstrated 100% sensitivity and 100% negative predictive value for malignancy. The positive predictive value was 84.2%, with specificity at 97.9%. Notably, all 32 histologically confirmed malignancies were correctly identified as suspicious endoscopically, though six lesions suspected of malignancy were histologically benign.
Yield of Biopsy in Normal Mucosa: Among the 140 patients with a macroscopically normal colonoscopy, 21.4% had abnormal histopathology. These abnormalities were almost exclusively non-specific colitis/inflammatory changes; no cases of malignancy, IBD, or TB were found in this subgroup.
Predictors of Malignancy: Multivariable analysis identified two independent predictors of histologically confirmed malignancy: the presence of a mass lesion on colonoscopy (OR 89.1) and age >50 years (OR 2.67). Per-rectal bleeding was not an independent predictor. A model based solely on pre-endoscopy clinical variables (age and bleeding) achieved an AUC of 0.736, while the full model including endoscopic findings achieved an AUC of 0.987.

Significance and Claims
The study concludes that colonoscopic impression shows substantial agreement with histopathology and possesses excellent diagnostic accuracy for detecting malignancy. The authors assert that the finding that approximately one-fifth of patients with macroscopically normal colonoscopies harbor clinically relevant inflammatory pathology supports the continued use of routine biopsy in symptomatic patients, even when the endoscopic appearance is unremarkable.

Regarding malignancy risk stratification, the paper claims that age over 50 and the presence of a visible mass lesion are independent predictors. The authors suggest that these factors, along with a pre-endoscopy clinical model (age and symptoms), may assist in prioritizing patients for urgent evaluation in settings where colonoscopy access is limited. The study emphasizes that while visual assessment is highly reliable for ruling out malignancy, histopathological confirmation remains essential even for visually obvious lesions, as a minority of such lesions prove benign.

Limitations
The authors acknowledge limitations including the single-center design, which may affect generalizability, and the retrospective categorization of free-text findings into standardized groups, which introduces potential for misclassification. Additionally, inter-observer variability in endoscopic and histopathological interpretation was not separately assessed.