Department of Pathology, Division of Cytopathology, University of Texas Medical Branch, Galveston, Texas
Corresponding author: Vicki J. Schnadig, MD, Department of Pathology, Division of Cytopathology, University of Texas Medical Branch, 301 University Boulevard, 9.328, Galveston, TX 77555; Fax: (409) 772-8437; email@example.com
We thank Gwen Baillargeon, MS, of the Department of Preventative Medicine and Community Health, Office of Biostatistics, University of Texas Medical Branch, Galveston, Texas for assistance with statistical data and analysis.
The current American College of Obstetricians and Gynecologists guidelines state that cervical cancer screening should begin at age 21 years, regardless of sexual or obstetric history. However, previous studies have demonstrated that there is a small but significant subset of high-risk adolescents with extensive sexual and obstetric history who harbor a significant squamous cervical lesion. The objective of the current study was to use histologic and demographic data from adolescents (aged <21 years) who received Papanicolaou (Pap) tests to determine whether they benefited from early cervical cancer screening.
Adolescent girls who had Pap tests between 2000 and 2010 were included in the study. Demographic data, including obstetric history, number of sexual partners, age of first coitus, age at first pregnancy, menarche, smoking history, and Chlamydia and syphilis infection, were analyzed for associations with levels of cervical dysplasia.
Of 56,785 adolescent Pap tests, 277 (0.5%) were diagnosed as high-grade squamous HSIL, and 56 of those Pap tests (20%) were from patients who had grade 3 cervical intraepithelial neoplasia (CIN-3) on subsequent biopsy and/or excision. One patient had microinvasive cervical carcinoma identified on loop electrosurgical excision procedure at age 27 years after an HSIL Pap test. Increased parity was associated significantly with higher rates of CIN-3.
The most common worldwide sexually transmitted disease is human papillomavirus (HPV). Several studies have indicated that sexually active young women in the United States have positive HPV test rates as high as 50% within 36 months of coitus, with rates as high as 57% for adolescents alone. Regression of detectable HPV infection is common in adolescents with low-grade lesions. Greater than 90% of cytologically diagnosed low-grade squamous intraepithelial lesions (LSIL) and histologically diagnosed grade 1 cervical intraepithelial neoplasia (CIN-1) revert to normal within 3 years. Sixty percent of CIN-2 lesions in adolescents will regress within 3 years. Progression rates of CIN-3 in adolescents have not been well established, but an incidence rate of 1 or 2 cases of cervical cancer per 100,000 adolescent girls ages 15 to 19 years has been reported and accounts for 0.1% of total cervical cancer cases.
The current 2010 American College of Obstetrics and Gynecologists (ACOG) recommendation (Committee Opinion no. 463) states that cervical cancer screening should begin at age 21 years, regardless of sexual or obstetric history. Exceptions to this guideline are sexually active adolescents with human immunodeficiency virus (HIV) and those who are otherwise immunocompromised. In current practice, many adolescents continue to have Papanicolaou (Pap) tests, and those with high-grade squamous intraepithelial lesions (HSIL) are referred for colposcopy and biopsy; and biopsy has been recommended for adolescents with persistent HSIL. Adolescents with biopsy-confirmed CIN-2 may be managed by observation with colposcopy and cytology. Neither the risk for progression of CIN-3 nor the recommended follow-up and treatment of CIN-3 in adolescents has been established. Excisional procedure has been recommended for adolescents who have biopsy-confirmed CIN-3. Conservative follow-up with Pap and colposcopy also may be an option given the rarity of invasive carcinoma in adolescents. The most recent guidelines obviate the associated angst regarding adolescent HSIL therapy by recommending that women aged <21 years not undergo cervical cancer screening.
The current cervical cancer screening guidelines are based on very low incidence rate of cervical cancer among adolescents, as stated above. Barnholtz-Sloan et al reported that rates of invasive cervical cancer (ICC) in young women ages 15 to 24 years were very low across all ethnic/racial lines with a peak incidence of ICC at ages ≥65 years. Those authors also stated that cervical cancer screening in adolescents has produced no evidence of reducing the rate of cervical cancer in this group. There is special concern about the potential for harmful follow-up treatment of abnormal Pap test results, including the risk of preterm delivery, preterm labor, and premature rupture of membranes. Although the ACOG recommendation to delay screening until age 21 years seems valid based on the very low incidence of ICC in adolescents, a small number of persistent CIN-3 lesions will go undetected, and guidelines suggest that there will be “catch-up” detection of these HSIL after the initiation of screening at age 21 years.
CIN-3 has been identified by biopsy in 7% to 8% of adolescents who were referred for colposcopy.[7, 8] Because the natural history of CIN-3 is unknown, the potential risks of allowing CIN-3 to remain undetected in adolescent women may continue to cause anxiety among health care professionals, at least until adequate data are acquired to confirm a high benefit-to-harm ratio and initiation of screening at age 21 years becomes the accepted norm.
Our institution serves a large population of sexually active adolescents with a high rate of teenage pregnancies. Until recently, these young women have undergone routine Pap screening. The objective of the current study was to determine whether there is a subgroup of adolescent girls who may benefit from Pap screening before age 21 years. Data that would argue for earlier screening in this subgroup would be a statistically higher incidence in biopsy-proven CIN-3 or the presence of ICC. Demographic data, biopsy results, and clinical outcomes were correlated with Pap test results to test for an increased risk of CIN-3 or ICC among certain subsets of adolescent women.
MATERIALS AND METHODS
A computer-based search of the years 2000 to 2010 was performed to identify Pap test results from adolescents, defined as females aged <21 years. Pap results were recorded and tabulated as negative, LSIL/atypical squamous cells of undetermined significance (LSIL/ASCUS), or HSIL.
A chart review of adolescents who had at least 1 HSIL Pap test was conducted to obtain results from cervical biopsy and loop electrosurgical excision procedure (LEEP), if performed, and patient demographic information, when available, including age of first coitus, number of sexual partners, obstetric history, sexually transmitted disease history, smoking history, and HIV status. Consistent and full demographic data were available only from patient records after 2005, when a standard template questionnaire was implemented in the electronic medical record (EMR). A sample of 48 adolescents who had negative Pap tests was used as a baseline for demographic comparison. Fisher exact tests were used to analyze associations between smoking history, Chlamydia, syphilis, and HIV infection and CIN level. A 1-way analysis of variance (ANOVA) procedure was used to assess whether the variation in each of the categories, including number of sexual partners, age of first coitus, and age of first pregnancy, was because of differences among levels of CIN. For variables gravidity and parity (categorized as 0, 1, and ≥2), a Kruskal-Wallis test was used for analysis. Data were analyzed using the SAS statistical software package (version 9.2; SAS Institute, Inc., Cary NC). Institutional Review Board approval was obtained for the retrospective study.
There were 56,785 adolescents (age range, 13-20 years) who had at least 1 Pap test during the study period. Of the 56,785 adolescent Pap tests, 45,276 tests (79.5%) were interpreted as negative for intraepithelial lesion or malignancy, 11,232 tests (20%) were interpreted as ASCUS or LSIL, and 277 tests (0.5%) were interpreted as HSIL. In the HSIL group, 89 patients (32%) did not undergo a biopsy. All patients who had an HSIL Pap result were recalled for colposcopy. Failure to undergo a cervical biopsy and/or endocervical curettage usually was because the patient failed to respond to a certified letter and was lost to follow-up. The breakdown of patients who underwent biopsies was as follows: 35 patients (13%) had no dysplasia on biopsy, 57 (21%) had CIN-1, 40 (14%) had CIN-2, and 56 (20%) had CIN-3. The age range for adolescents with CIN-3 was 14 to 20 years. These data are summarized in Table 1. One patient had microinvasive cervical carcinoma identified on subsequent LEEP. This patient was aged 20 years at the time of the initial HSIL Pap result; was lost to follow-up between 2000 and 2004; had negative Pap results in 2004, 2005 (pregnant), and 2006 (postpartum); and had ASCUS and atypical squamous cells/cannot rule out HSIL results in 2007. A biopsy in 2007 revealed CIN-3, and focal microinvasion (<1 mm) was identified by LEEP (1 Figure 1). Subsequently, this patient underwent a second LEEP, on which CIN-1 was identified. She has had negative Pap test follow-up through 2011.
Table 1. Breakdown of Papanicolaou Test Interpretations and Histologic Outcomes
The means, standard deviations, and ranges of demographic data, sorted by levels of dysplasia, are summarized in Table 2. The Pap test-negative, CIN-1, CIN-2, and CIN-3 groups were followed for an average of 41.2 months, 38.4 months, 55.2 months, and 43.8 months, respectively. The sample numbers in Table 2 differ from those in Table 1 because of inconsistent and missing demographic information in the medical record. Only patients who had some of the demographic data available were included in Table 2. Data were obtained more consistently in recent years, during which patient-reported information was entered into a standard template in the clinics. The percentages of patients with each data element available are provided in Table 2.
Abbreviations: CIN-1, cervical intraepithelial neoplasia grades 1; NILM, negative for intraepithelial lesion or malignancy; Pap, Papanicolaou test; SD, standard deviation.
aNumbers include only patients for whom demographic data could be retrieved from the medical record; thus, the numbers are lower than those provided in Table 1.
NILM on Pap test: Controls, no biopsy performed
Age at menarche, y
Age of first coitus, y
Age of first pregnancy
No. of sexual partners
0.56± 0.71 [0-2]
Age at menarche, y
Age of first coitus, y
Age of first pregnancy, y
No. of sexual partners
Age at menarche, y
Age of first coitus, y
Age of first pregnancy, y
No. of sexual partners
Age at menarche, y
Age of first coitus, y
Age of first pregnancy, y
No. of sexual partners
We obtained information regarding history of smoking, HIV status, Chlamydia infection, and syphilis for 117 patients and used the Fisher exact t test to assess for differences between these conditions and increasing levels of CIN. No significant association was observed between these conditions and increasing levels of CIN; however, a tendency toward significance was observed for Chlamydia infection and syphilis. These results are provided in Table 3.
Table 3. Demographic Correlation With Different Degrees of Cervical Intraepithelial Neoplasia
The Kruskal-Wallis test result for gravidity revealed no significant differences between CIN-1, CIN-2, and CIN-3. ANOVA of the number of sexual partners, age of first coitus, and age of first pregnancy indicated that there were no significant differences between CIN levels; although, again, there was a trend toward significance for age of first coitus and age at the time of first pregnancy. The Kruskal-Wallis test revealed a statistically significant increase in CIN level as parity increased. Missing data and sample size, of course, may affect any significant findings. A summary of these statistical test results is provided in Table 3.
We believe that the best strategy for the prevention of cervical cancer in adolescents is the implementation of universal HPV vaccination programs, which, when applied before the onset of sexual activity, have been demonstrated to be highly efficient and cost-effective.[9, 10] However, compliance with the HPV vaccination regimen among eligible adolescents in the United States has been disappointing. For the recommended treatment with 3 doses of HPV vaccine, the rate of coverage for adolescent girls in the United States increased from 26.7% in 2009 to only 32% in 2010, and the percentage of girls who received 1 dose of HPV vaccine increased from 44.3% in 2009 to only 48.7% in the same period. Thus, there is a significant percentage of unvaccinated adolescent girls who are susceptible to high-risk HPV infections. Given this low rate of compliance with vaccination programs and the presence in some populations of high rates of early onset sexual activity and teen pregnancy, clinicians may question whether delaying Pap testing until age 21 years may result in failure to diagnose and treat serious conditions. Of course, the question should not be whether CIN-3 would go untreated in adolescence or even whether a very small number of early invasive cancers would present before the onset of screening, but whether this delay actually will result in significant patient morbidity or loss of life. Concerns over decreases in test-generated income must not enter into the rationale for limiting Pap or HPV testing. There are 2 questions addressed in our study: 1) Was significant morbidity or life-threatening disease identified in our patient group? The answer is no! 2) Does certain demographic information exist that places a young woman at higher risk for early onset ICC? We believe that this second question merits more systematic data collection.
In our current study, of 56,785 adolescents who receiving Pap tests, 277 had HSIL, for a rate of 0.5%. The low incidence of HSIL is consistent with previous studies[12, 13] and confirms that many adolescents would need to be screened to identify 1 case of HSIL. We identified no adolescents with invasive carcinoma and 1 adolescent with microinvasive cervical carcinoma, which was discovered during postadolescent follow-up. This patient had her initial Pap at age 20 years, near the new guideline age for the onset of testing, and actually had an interval of several years with negative cervical cytology, followed by ASCUS, and finally atypical squamous cells/cannot rule out HSIL. These findings may indicate either regression of the original lesion with emergence of a second lesion or the presence of an occult lesion that was missed in some of her Pap tests. This patient underwent two therapeutic LEEPs and has had benign follow-up. These findings indicate that failure to screen in adolescence would not have adversely affected the patient's outcome.
An unexpected bit of information emerged from our study. The rate of adolescents with HSIL Pap results for whom no follow-up data were available was well in excess of the rate for our general population. Over the years, despite having a population composed largely of low-income patients and prisoners, our biopsy rates for patients with HSIL have consistently been >80%. Efforts to contact patients by both telephone and certified letter are part of our clinic's protocol for follow-up of all HSIL results; and, unless patients refuse follow-up, prisoners with HSIL are routinely sent to our gynecology clinic. Biopsy rates for prisoners with HSIL in 2010 and 2011 were 76% and 89%, respectively. This unexpected no-biopsy rate may represent a less compliant attitude toward follow-up in this age group and indicates another potential expense-without-benefit aspect of screening in this population.
Study findings support the ACOG guideline to begin cervical cancer screening at age 21 years. Nonetheless, our study indicated a follow-up biopsy or LEEP excision confirmation of CIN-3 in 56 of the 277 adolescents (20%) who had an HSIL Pap result. Other studies have documented a small but significant subset of adolescents with CIN-3 or greater lesions. Three studies reported that 24 of 335 adolescents (7.2%), 15 of 131 adolescents (7.8%), and 13 of 50 adolescents (26%) who had an HSIL diagnosis had CIN-3 on a follow-up biopsy.[7, 12, 14] Clements et al studied 3137 adolescents who were referred for colposcopy and reported that 48 (8%) had CIN-3. The differing rates of CIN-3 in adolescents are most likely because of differences in patient populations, and our high rate of CIN-3 on follow-up of HSIL may reflect factors unique to our adolescent population. Adolescent populations with higher incidences of CIN-3 may require modification of current ACOG screening guidelines if evidence emerges that these populations have an increased risk for early onset ICC. Our data do not support a need to modify these guidelines.
Despite the evidence for high-grade CIN lesions in adolescents, the risk of invasive carcinoma during adolescence is very low.[7, 13, 14] The Surveillance, Epidemiology, and End Results Program of the National Cancer Institute reported that, from 1998 to 2006, 14 cervical cancers per year were identified in girls ages 15 to 19 years, translating to an incidence of 0.1 per 100,000. The one microinvasive cervical carcinoma in our study occurred when the patient was aged 27 years, and this patient's lesion would have been detected and treated under the current screening guidelines. The low rate of cervical carcinoma in adolescents, in part, is because of the high regression rates of some high-grade lesions (CIN-2), but the treatment of screen-detected CIN-3 also may contribute to very low cancer rates. The natural history of adolescent CIN-3 is unknown, possibly because the treatment of screen-detected CIN-3 may preclude knowledge of its rate of progression to ICC. However, some indirect data are emerging from Great Britain, where the age for the onset of cervical carcinoma screening changed from age 20 years to age 25 years in 2004, and a 10.3% increase in cervical cancer incidence was observed among women ages 20 to 29 years between 2000 and 2009. An increase in cervical cancer also was reported for women ages 20 to 24 years, with an increase that tended toward significance. It is noteworthy that Patel et al suggest that this increase likely is not related to new screening guidelines but, rather, is related to exposure to predisposing factors, and those authors stress the need to assure screening coverage for women ages 25 to 29 years. Again, before we malign delayed onset of screening, we must determine whether the delay in diagnosis of these patients with early onset ICC actually results in significant morbidity or loss of life and whether early screening would have prevented the ICC. In our case of microinvasive carcinoma, the patient had 1 HSIL Pap result followed by several years of benign follow-up and pregnancies before she developed a curable microinvasive carcinoma.
Another study that monitored all types of cancer in Malawi, Africa, reported that 574 of 4787 cervical cancers (12%) occurred in the group ages 20 to 29 years. This is noteworthy, because that country's screen and treat program specifically targeted patients aged ≥30 years. A Swedish study indicated that women who failed to obtain a Pap test within the recommended screening interval were at higher risk for cervical cancer than women who had Pap testing. This increased risk for cervical cancer is present for all age groups, including the group ages 21 to 29 years. These studies suggest that a small but significant subset of young women are at risk for cervical cancer, stressing the importance of screening for women in young adulthood. The ACOG guideline to begin Pap testing at age 21 years is appropriate, but the medical community needs to have safeguards in place to assure that screening and follow-up of abnormalities are initiated at the age set by the guidelines.
It would be interesting to acquire a large, multi-institutional database that contains consistently obtained and accurate information regarding patient demographics other than ethnic and racial origins. Ethnicity and race, in fact, often are ill defined and of limited value, because neither, as recorded in medical records, necessarily correlates with genetic factors or socioeconomic status. Thus, we chose to focus on demographic information related to individual patient behavior. We believe that it would be valuable to develop a universal database with acquisition of very large amounts of data to determine whether there are certain risk factors that would place adolescents at increased risk for early onset CIN-3 and early progression to ICC. Template questionnaires that have been integrated into the electronic records of patients in the gynecology clinics at our institution provide a standardized record of demographic information that is useful for assessing risk factors for CIN-3. These templates allow for standardized collection of data regarding age at menarche, age of first coitus, number of sexual partners, pregnancy history, history of sexual violence, sexually transmitted diseases, etc. Routine input of data into these templates has the potential to provide important information for assessing the risk of early onset CIN-3 by multivariable analysis. This may allow the categorization of adolescents into high-risk and low-risk groups and the selection of those who may benefit from the early initiation of cervical screening. Our current study was limited by the availability of template-generated data only for patients who attended clinics during or after 2005.
Only a few studies have addressed the association of demographic variables with grades of cervical dysplasia in adolescents. In a multivariable analysis, Clements et al observed that adolescents who had ≥4 lifetime sexual partners had a significantly increased risk of CIN-3 versus CIN-1 or less lesions, with an adjusted odds ratio of 2.11 compared with those who had <4 sexual partners. Another study that included both adolescents and young women (ages 13-24 years) had a 36% higher risk of CIN-3 versus CIN-1 for each additional year of oral contraceptive pill use; however, the P value used in that study was 0.1. Those authors suggested epigenetic phenomena to explain the increased risk. The risk factors in the current study that were associated with CIN-3 or worse lesions have not been observed consistently in other studies. The results from these 2 studies demonstrate promise for the use of demographic data to categorize adolescents who have the greatest risk of developing significant cervical lesions.
In our study, we observed that parity differed significantly between CIN levels, with a statistically significant increase in CIN grade with increasing number of births. Increased risk of high-grade cervical lesions in pregnancy is not an established finding. It is noteworthy that higher levels of CIN with increasing number of births also was reported by Clements et al, who observed a significant difference in parity between CIN-3 and CIN-1 or less lesions (P = .02) in univariate analysis. However, that finding was not significant in a multivariate analysis. The preponderance of evidence suggests that HPV during pregnancy does not appear to promote progression of cervical lesions. Proponents of delayed screening argue that, although CIN-3 is present in a subset of adolescents who have multiple children, the slowly progressive nature of CIN-3 to invasive cancer, averaging between 8.1 and 12.6 years, will negate any life-threatening consequences of delaying screening to age 21 years.
A definite limitation of this study is the significant number of missing demographic data points; as such, any conclusions gleaned from the demographic data are subject to question. However, it is important to note that the integration of standardized questionnaires into the EMR may allow for better data collection for future investigations. Parity was the most readily available data in our cohort, which explains the greater significance of the findings related to it. An interesting possibility, which would require further data collection and investigation, is whether young women who carry more pregnancies to term (and consequently have a longer period of relative immunosuppression) are at higher risk for CIN-3 than young women who do not carry their infants to term. Thus, as the EMR is used for a longer time, more complete data will be collected, although data collection on adolescents and adolescent Pap test results may cease with the new guidelines, and data collected from women aged ≥21 years must be used. Studies will monitor the incidence of ICC among women between ages 21 and 25 years and will continue to correlate the early onset of ICC with demographic factors.
Another guideline change to be considered along with screening initiation at age 21 years is the proposed increased screening interval from yearly to every 2 or 3 years for women between ages 21 and 30 years who have had a negative Pap. HPV screening tests are not indicated for this group. These 2 guidelines may increase worry that a few of these high-risk adolescents may fall through the system because of a solitary false-negative screening Pap. Only time will tell whether this increase in the screening interval will result in harm generated either by false-negatives or by an increase in reports of ASCUS because of pathologists' fears of missing significant lesions.
In summary, there is a small but significant subset of adolescents who harbor CIN-3. Whether these CIN-3 lesions require definitive therapy in adolescence still may be open to question given the rarity of ICC before age 25 years. Our study failed to identify any patients who presented ICC before age 25 years, essentially supporting the concept that delaying cervical screening until age 21 years does not put young women at risk for increased morbidity or mortality. However, there may be some demographic traits that place young women at increased risk for developing CIN-3 in adolescence, but this lesion may not place women at increased risk for developing advanced ICC before the onset of screening at age 21 years. Although we concur that screening the majority of adolescent women has the potential to do more harm than good, we advocate ongoing systematic studies of patient demographics, potentially by using standard EMR-generated questionnaires, to assess the existence of a unique subgroup of adolescents who may benefit from early screening.