Skip to Main Content
Principal Investigator
Thomas O'Grady
University at Albany School of Public Health
Position Title
Doctoral Student, Teaching Assistant
About this CDAS Project
PLCO (Learn more about this study)
Project ID
Initial CDAS Request Approval
Aug 1, 2013
Assessing the Effect of Increased Medical Surveillance on Thyroid Cancer Incidence in the Prostate Lung Colorectal & Ovarian Cancer Screening Trial and the National Lung Screening Trial: A Pooled Study of Two Randomized Control Trials
Thyroid cancer incidence has been increasing in recent years; however, the disease is characterized by a low mortality rate. Established risk factors for thyroid cancer include exposure to ionizing radiation, especially in childhood, as well as past history of thyroid nodules or thyroid disorders. Recent reports indicate that other factors which increase thyroid cancer risk are obesity and excess weight gain in adulthood, and dietary nitrate/nitrite intake, while factors which likely decrease thyroid cancer risk are fruit and vegetable consumption.

A theory frequently mentioned in the literature regarding the trends seen in thyroid cancer is that increased medical scrutiny has led to the finding of incidental thyroid carcinomas that would never affect the health of the patient. For this hypothesis to be true there would need to be a large sub clinical reservoir of thyroid tumors present in the general population, which is supported by autopsy studies.

Using the National Cancer Institute’s Surveillance Epidemiology and End Results (SEER) data it has been shown that tumors of less than 1 cm account for approximately half the increase seen in papillary thyroid cancer from 1988 to 2002 while tumors of less than 2 cm accounted for 87% of the increases seen in papillary thyroid cancer during the same time period. Additionally SEER data has been used to demonstrate a 19% annual increase in papillary thyroid carcinomas from 1985-2006 which outpaces increases in larger papillary thyroid carcinomas by 7-9% per year. Data from the Ontario Cancer Registry indicates the incidence of small thyroid tumors is outpacing that of larger tumors and that a significant yearly increase in thyroid cancer matched a steady increase in number of per capita CT scans. Registry data from Wisconsin and New Jersey show that areas with the highest socioeconomic status, highest education, and greatest access to healthcare have the highest rates of thyroid cancer. A study in New South Wales, Australia studied the pathway to diagnosis in 452 thyroid cancer patients to evaluate the impact of medical surveillance and increased technological sensitivity on the detection of asymptomatic disease. The report found that roughly 60% of thyroid cancer diagnosis was incidentally detected during a medical encounter.

Evidence from the previous paragraphs indicates the importance of conducting a study to investigate in greater detail the effects that clinical practice has on detecting thyroid carcinomas in the population. The Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial and the National Lung Screening Trial (NLST) are great candidates to carry out further research of this kind.

(Citation listing does not fit in this window. A listing of citations used in preparing this summary is available upon request of the investigators.)

We propose to examine the following specific aims using data from the PLCO and the NLST studies. The first specific aim is to assess the incidence rates of thyroid cancer for individuals randomized to each arm of the trials. Second, we will assess whether the incidence rate in each arm of the trial (for PLCO) varies with times. A potential reason for a varying incidence rate is that screening was carried out in the active arm primarily in the first six years. It is possible that the incidence rate for thyroid cancer was higher during this period of time and then leveled off or decreased. If this is the case we would hypothesis a consistent incidence rate in the control arm throughout the trial. Third, we propose to pool both the PLCO trial and the NLST to determine the combine effect of higher levels of medical surveillance on thyroid cancer incidence. While previous studies using SEER data were valuable in forming hypotheses about thyroid cancer trends their primary limitation was the descriptive nature of the studies. The PLCO trial and NLST offers a unique opportunity to directly assess the effect of randomization to increased medical attention on thyroid cancer outcomes. Additionally biasing factors such as confounding should be minimized due to the large sample size and the randomized nature of the trial.

The reason for studying these specific aims is twofold. The first is that through additional medical encounters those randomized to the screening arm had more opportunities for a clinician to find an incidental thyroid carcinoma. The second is the possibly that any chest x-rays received during the PLCO trial may have imaged enough of the neck so that incidental thyroid tumors were detected with a greater frequency by imaging. Lastly, in the NSLT low dose helical CT scans are much more sensitive to picking up incidental nodules than standard x-rays. It is our belief that using the PLCO trial and NLST to accomplish our specific aims will give a better understanding of the types of medical encounters which lead to incidental thyroid carcinomas.

There were approximately 250 incident thyroid cancer cases diagnosed in the PLCO Cancer Screening Trial between 1993 and 2009 and approximately 60 incident thyroid cancers diagnosed in the NLST between 2002 and 2009. Hazard ratios will be calculated for thyroid cancer incidence and time to thyroid cancer diagnosis using Cox proportional hazard models. We will use competing risk analysis to assess whether the incidence rate of thyroid cancer is changing in relation to other outcomes during the course of each trial. We will also calculate hazard ratios separately for the papillary and follicular histological subtypes. Age, body mass index, cigarette smoking, and gender will be considered as potential confounding and effect modifying variables. A log linear model will be constructed to compare grade of thyroid cancer for individuals randomized to screening versus those randomized to standard care.


Thomas J. O’Grady Investigator UAlbany

Francis P. Boscoe Investigator UAlbany/NYSDOH

A. Gregory DiRienzo Investigator UAlbany

Margaret A. Gates Investigator UAlbany

Cari M. Kitahara Investigator NCI, DCEG, NEB

Related Publications