Scientific Paper Critique

Article: Canine scent detection in the diagnosis of lung cancer: revisiting a puzzling phenomenon

Authors: R. Ehmann, E. Boedeker, U. Friedrich, J. Sagert, J. Dippon, G. Friedel and T. Walles

 Introduction

           

Lung cancer is the second most common cancer and the most common cause of death from cancer. While early detection is key, issues arise in that its early stages are asymptomatic and there is currently no screening method. As well, modern techniques are not ideal for discriminating between healthy and lung cancer patients and so a better method of detection, preferably one that can detect the cancer early on is desired.

Most cancers are associated with volatile organic compounds (VOC's). These are compounds that patients breath out during the early stages of cancer. “Electronic noses” are the newest technology used to detect these however there are limitations to this procedure including very sensitive instruments, the patients cannot smoke or eat before the procedure, it takes a long time to get results, etc. As well, there have been no VOC’s identified for lung cancer.

Methods

           

The researchers obtained ethical permission and collected breath samples of healthy individuals (Group A), lung cancer patients (Group B) and chronic obstructive pulmonary disease patients (Group C) from volunteers who provided signed consent. Their medical history, including medications they were taking, was recorded to control for any confounding variables. The breath samples were taken by the volunteer breathing into a cylindrical glass tube, lined with polypropylene fleece, five times. The tube was sealed and handled by all individuals to avoid “scent labeling”.

            

Four house dogs were trained in a separate room dedicated solely to said task and were taught to lie on the floor with their nose on the tube when cancer was detected.

            

Three tests were performed within two days:

1)   Identifying healthy controls from lung cancer patients

2)   Identifying lung cancer from COPD

3)   A mixed study of COPD and healthy patients

Two people observed the dogs and were blinded to avoid bias. Following this a variety of statistical tests on the data were performed.

Results:

           

The official number of volunteers were 110 healthy individuals, 60 lung cancer patients and 50 COPD patients. The researchers found that the overall sensitivity of the dogs detection was 71% and the specificity was 93%. The dogs had the greatest success in the mixed study (COPD and healthy patients) and had the least success in identifying lung cancer versus healthy controls. They also discovered that the dogs were not more accurate in identifying advanced stages tumors and in fact the highest accuracy was for stage I tumors (100%) and the lowest for stage VI (63%). From this data they concluded that there must be some kind of stable, detectable marker found in the breath of lung cancer patients which is independent from that of COPD. As well, this marker must be present in the initial stages of lung cancer and thus it may prove to be a good, non-invasive screening test for lung cancer. Further work needs to be done to determine what specific compound this VOC is.

Personal Critique:

            I found this article to be very interesting. It is amazing to think that dogs have such an acute sense of smell to be able to detect cancer and that they may be used in medicine.

            

The authors did a fantastic job, in my opinion, of explaining their research and findings. Step by step descriptions of each aspect of the experiment was given- from the initial collection of patients, to training the dogs, analyzing the findings, accounting for bias and confounding variables and discussing the results and their implications. They also included many headings to divide the article into sections which made the experiment easy to follow and understand.

           

I found the researchers to be very thorough in their experiment and it seems as though they tried to account for all possible variables. They were sure to obtain signed consent from the patients and collected their medical histories and medications to prevent confounding variables. 

They then performed statistical tests on any possible confounders to determine if they influenced the results.  In addition, they handled their instruments with care- being sure not to introduce any outside scents that may affect the results.  They used multiple statistical models to evaluate their results (these could have been explained in more detail) to ensure accuracy. Their results were then presented clearly- both in words, figures and tables- and conclusions clearly drawn. I felt as though all the necessary information and details were included to give a comprehensive picture of the experiment.

One thing that I believe could have been improved would be to obtain patients from different areas. The researchers used volunteers from only two facilities in Germany so the questions of generalizability and bias can be raised.

Overall I found this article very informative, well conducted and written. It has a very wide audience since lung cancer is such a common disease and I believe it was written in such a way that many people could read and understand it. A very important aspect of this article is that it provides direction for future research- research must now be conducted to determine what the specific VOC found in lung cancer patients is. 

Article Citation:

Ehmann, R., Boedeker, E., Friedrich, U., Sagert, J., Dippon, J., Friedel, G., & Walles, T. (2012). Canine scent detection in the diagnosis of lung cancer: revisiting a puzzling phenomenon. European Respiratory Journal, 39(3), 669-676.