Our original work in China was a cross-sectional comparative trial that, for the first time, directly measured sensitivity and specificity of five cervical cancer screening technologies. In October of 1998 we conducted a pilot study in Xiangyuan and Yangcheng Counties in Southwest Shanxi Province, Peoples Republic of China. We demonstrated a high prevalence of cervical neoplasia as well as the feasibility of a large cross-sectional comparative trial of multiple screening techniques. A report of this pilot trial was published in the International Journal of Gynecologic Cancer, 9:411-417, 1999. In the summer of 1999, a full clinical trial was completed (The Shanxi Province Cervical Cancer Screening Study – SPOCCS I). Two thousand women between 35 and 45 years of age from Xiangyuan County were recruited by a field supervisor from the Department of Epidemiology of the Cancer Institute/Hospital of the Chinese Academy of Medical Sciences in Beijing and by the local commune nurse (much like our LPN’S, often referred to as “Barefoot Doctors”). The fact that at least five cervical biopsies were obtained on every participant in the trial using special biopsy instruments (even those who had negative screening tests and negative colposcopy ) allowed us to directly measure the sensitivity and specificity of the screening tests. (POI micro-biopsy protocol )
Data from the full screening study performed in the summer of 1999 was published in Gynecologic Oncology 83: 439-44, Nov 2001 . In April of 2000 our research team received the first prize award for our work at Eurogin 2000 in Paris.
We initially began our work in China because the rates of cervical cancer in rural China were reported to be very high. The Shanxi Province Cervical Cancer Screening Study (SPOCCS) was designed to determine the sensitivity and specificity (critical in low resource settings) of five screening technologies in order to develop low-cost screening for rural China. We searched for a screening algorithm that was sensitive for cervical intraepithelial neoplasia (CIN) II, III, and cancer, and highly specific for CIN II and III. That search continues to this day, but much progress has been made. With such an algorithm, it would be possible to ablate the transformation zone of the cervix in test positive subjects without histological confirmation of the diagnosis. In SPOCCS I we examined 1997 women ages 35-45. Each subject underwent a self-test for HPV (by HC-II assay), fluorescence spectroscopy, liquid based cytology (ThinPrep)(read manually and used as a direct test for HPV), a visual inspection (VIA) diagnosis, and colposcopy with multiple cervical biopsies (see chart below). Our biopsy protocol serves as the critical element in our studies because it results in data with minimal verification bias. The POI micro-biopsy protocol that we have always used has been central to our work.
Test Sensitivity and Specificity – SPOCCS I
|# PositiveSensitivity for ³CIN IISpecificity for ³CIN IISensitivity for ³CIN IIISensitivity for Cancer|
|HPV Self-Test( ³1.0 pg/ml)||17%340/1,997||83%71/86||86%||81%35/43||75%9/12|
|Fluorescence Spectroscopy **||91%1,759/1,934||94%81/86||9%||93%40/43||100%12/12|
|ThinPrep®Pap ( ³ASCUS)||25%506/1,997||94%81/86||78%||98%42/43||100%12/12|
|HPV Direct Test( ³ 1.0 pg/ml)||18%364/1,997||95%82/86||85%||98%42/43||100%12/12|
|Visual Inspection(Any abnormal)||28%552/1,997||71%61/86||74%||79%34/43||67%8/12|
In SPOCCS I (above) a fluorescence spectroscopy** device that worked well in the esophagus was unable to overcome the cervical inflammation that was not present in U.S. testing, but is present so often in areas of the world with high rates of cervical cancer. This is an excellent example of the importance of not only tissue specific, but environment specific clinical validation.
Multiple manuscripts on visual inspection techniques for screening, HPV testing, self-sampling, accuracy of colposcopy, and liquid based pap test technology have been published by POI using SPOCCS I data. Six year follow-up studies have also been published that explored the durability (negative predictive value) of a negative HPV test as well as the significance of HPV “titer” both at the time of the primary screen and the follow-up exam. On the basis of the data from our work in SPOCCS , we began a 9000 patient trial in rural China (SPOCCS II) studying a new bio-molecular marker for diagnosing pre-invasive disease.In addition we have developed, tested and published a survey instrument that explores cross-cultural differences in women’s acceptance of self-sampling technology.
SPOCCS II took place between October/2000 and April/2002. 8,497 women age 27 to 56 years were screened for cervical neoplasia with self test using a brush (rather than the swab as in SPOCCS I) and the same brush for a direct test for high-risk HPV DNA (HC II). Liquid -based cytology used Sure-Path (in SPOCCS I we used ThinPrep). Again our biopsy protocol was used to ascertain the true diagnosis.
The SPOCCS I and II studies generated many important findings that have been published in the medical literature. Theses studies upon completion of these two trials contain more than 20,000 matched colposcopic impressions and biopsies. In addition to the patient care that has been provided, one of the most important contributions from the SPOCCS studies is the clear documentation of the inaccuracy of colposcopy.
The colposcope is the most commonly used imaging technology for the diagnosis of pre-invasive and invasive cancer of the cervix. These studies gives further strength to the importance of our biopsy protocol in the validation of new technology as well as in patient care.
Since colposcopy directed biopsy has been the standard for the confirmation of disease for more than 40 years, there are a multitude of published clinical studies whose reported results suffer from profound verification bias. Recently we re-analyzed the biopsies from SPOCCS I using a average epithelial thickness (measured perpendicular to the epithelium with a micrometer) and the average nuclear density, to morphologically characterize our biopsies. We have determined that there are high grade lesions than are not visible with colposcopy (by virtue of being very thin lesions) and therefore responsible for a significantly large portion of false negative colposcopic exams. In addition there are benign thick lesions that are false positive diagnoses. Several manuscripts have been published documenting the inaccuracy of colpo-directed biopsy, the causes, and the solution.