Correspondence |

Great Data, But How Do We Use It?Response FREE TO VIEW

Frank Detterbeck, MD, FCCP
Author and Funding Information

Affiliations: Yale University, New Haven, CT,  Graduate School of Medical Sciences, Kumamoto University, Kumamoto, Japan

Correspondence to: Frank Detterbeck, MD, FCCP, Department of Surgery, Yale University, FMB 128, 330 Cedar St, New Haven, CT 06520-8062; e-mail: frank.detterbeck@yale.edu

Chest. 2008;133(3):827-828. doi:10.1378/chest.07-2489
Text Size: A A A
Published online

I read with interest the article by Ikeda et al,1 and I commend the authors for their ongoing work to defining objectively how to manage patients with small lesions. This is an important area, if we are to avoid mistakes of leaving significant cancers unresected and at the same time avoid risks and morbidity from unnecessary biopsies in patients with nonthreatening lesions. The authors are clearly leaders in the field.

Having said this, I have trouble with an actual application of their results. Although the authors have focused on defining sensitivity and specificity, application of test results to an individual patient requires knowledge of the false-negative and false-positive rates, which cannot be estimated from the sensitivity and specificity. These rates are not sufficiently low for the parameters defined, in my opinion, to allow application.

Let us first assume that we intend to avoid biopsy of atypical alveolar hyperplasia (AAH) but think that it is important to resect a bronchioloalveolar carcinoma (BAC) or an adenocarcinoma. The first test cited was whether there were one or two peaks on the histogram. The data indicate that if there is one peak, then 33% of the lesions were AAH, but 67% were either BAC or adenocarcinoma. Thus, resection would be needed. If there are two peaks, none of the lesions are AAH (all are BAC or adenocarcinoma), an indication that resection would be needed regardless of whether there are one or two peaks on the histogram.

Another parameter to differentiate between AAH and BAC is the 75th percentile CT number. If this number is ≤ − 584 Hounsfield units (HU), then 36% of the lesions are BAC, indicating that resection would be needed even though the majority are AAH. If the number is > − 584 HU, then 94% of the lesions are BAC. So once again, resection is needed either way.

The mean CT number comes into play if the goal is to resect only adenocarcinoma and observe BAC (with the assumption this is an indolent tumor for which careful observation is justified). If the mean CT number is > − 472 HU, then 69% of lesions are adenocarcinoma and 31% are BAC. If the mean CT number is ≤ − 472 HU, then 15% of lesions are adenocarcinoma. Although 85% of lesions are BAC in this case, the rate of adenocarcinoma is too high in my opinion to justify observation.

Perhaps I have misunderstood how the authors propose to use the parameters they have defined. I raise these points not as a criticism but in searching for clarification. I think the problem arises from trying to go from sensitivity, specificity, and accuracy to definition of the reliability of a test result in an individual patient in order to feel comfortable basing clinical decisions on this particular test result. I do very much, however, appreciate their efforts to provide objectivity in management of a difficult clinical issue.

Finally, I wonder if the real answer to the dilemma of how to manage these patients will come when we are able to define where patients lie along a continuum instead of assigning them to histologic categories. AAH appears to be a precursor for BAC and adenocarcinoma; however, at least some BACs and adenocarcinomas are indolent tumors, which may take 10 or 20 years to develop into a large enough tumor to be lethal. A better understanding of these lesions as part of a continuum would help balance the need for intervention with the patient’s age and comorbidities.

The author has no conflict of interest to disclose.

The authors have no conflicts of interest to disclose.

Ikeda, K, Awai, K, Mori, T, et al (2007) Differential diagnosis of ground-glass opacity nodules.Chest132,984-990
To the Editor:

Dr. Detterbeck has pointed out that there were some false-positive and false-negative results with our procedure,1 which might cause an incorrect treatment plan. While we have no objection to his comments, our data just show the accuracy of differentiation among atypical adenomatous hyperplasia (AAH), bronchioloalveolar carcinoma (BAC), and adenocarcinoma (AD) by using CT number analysis. While visual analysis can hardly differentiate the nodules, our CT number analysis can evaluate the ground-glass opacity (GGO) nodules with objectivity. If the nodules showed the “two-peak pattern” on CT number histogram, we can deny AAH. Even if BAC had a 75% percentile CT number that was lower than − 584 Hounsfield units (HU), the malignant grade should be lower than those for which the 75% percentile CT number was higher than − 584 HU. Therefore, those BACs with a low 75% percentile CT number could be observed by CT, which would not miss the chance of curative treatment after follow-up. Dr. Detterbeck pointed out that 15% of ADs and 31% of BACs could be incorrectly classified with mean CT number, causing incorrect surgical procedures. However, in Japan, the final indication of wedge resection for BAC is usually determined by intraoperative frozen section. If the intraoperative frozen section showed AD, the operation is usually converted to segmentectomy or lobectomy. Furthermore, while 31% of BACs could be classified as AD with mean CT number, those BACs with high mean CT number could have higher malignant grade than those with a low mean CT number. Therefore, those BACs with a high mean CT number might be better to be treated with segmentectomy than wedge resection.

Radiologic imaging is not a final diagnosis. However, CT number analysis for GGO lesions could predict the histologic type more correctly than visual analysis. We believe that further advances in software will make the procedure used in our study more popular. We appreciate the valuable comments by Dr. Detterbeck.

Ikeda, K, Awai, K, Mori, T, et al Differential diagnosis of ground-glass opacity nodules: CT number analysis by three-dimensional computerized quantification.Chest2007;132,984-990




Ikeda, K, Awai, K, Mori, T, et al (2007) Differential diagnosis of ground-glass opacity nodules.Chest132,984-990
Ikeda, K, Awai, K, Mori, T, et al Differential diagnosis of ground-glass opacity nodules: CT number analysis by three-dimensional computerized quantification.Chest2007;132,984-990
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

CHEST Journal Articles
Differential Diagnosis of Ground-Glass Opacity Nodules*: CT Number Analysis by Three-Dimensional Computerized Quantification
PubMed Articles
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543