The introduction of liquid-based Papanicolaou (Pap) tests (LBPTs) has reduced the incidence of unsatisfactory Pap tests (UPTs), but little is known about their causes and significance, especially in the case of SurePath LBPTs.
The introduction of liquid-based Papanicolaou (Pap) tests (LBPTs) has reduced the incidence of unsatisfactory Pap tests (UPTs), but little is known about their causes and significance, especially in the case of SurePath LBPTs.
All unsatisfactory LBPTs from January 1, 2003 to December 31, 2006 were retrieved. The characteristics of patients, providers, and LBPTs; the reason for UPTs; and any cytologic or histologic follow-up within 24 months were recorded. Negative Pap tests that were evaluated immediately after a UPT served as a control group.
Of 243,006 Pap tests (95.5% SurePath LBPTs), 0.23% were unsatisfactory. Scant cellularity was the primary cause of SurePath UPT. Women in this UPT group were older, had more diagnostic Pap tests taken, less frequently were taking contraceptives or were pregnant, and were more likely to be menopausal or posthysterectomy. The 278 women who had UPTs had significantly higher rates of follow-up Pap tests (65.1% vs 22.2%), abnormal Pap tests (5.4% vs 1.4%), biopsies (10% vs 1%), and abnormal biopsies (5% vs 1%) than the 284 women in the control group, including 7 women with cervical intraepithelial neoplasia 1 (CIN-1), 1 woman with CIN-2, 4 women with CIN-3, and 2 women with endometrial hyperplasia. The UPT rates varied little between provider groups (physicians vs nonphysicians and gynecologists vs nongynecologists).
The frequency of UPTs in a predominantly SurePath LBPT-screened population was very low and was caused mainly by low cellularity. Similar to conventional Pap smears, unsatisfactory SurePath LBPTs had a higher risk of significant histologic abnormalities on follow-up than negative satisfactory Pap tests and could have benefited from a repeat Pap test or other evaluation, according to current management guidelines. Cancer (Cancer Cytopathol) 2009. © 2009 American Cancer Society.
The cervical cancer screening system using Papanicolaou (Pap) tests (PTs) is immensely successful, as demonstrated by the decrease in cervical cancer mortality rates in countries that have implemented organized screening programs.1 Most failures of cervical cancer screening are because of inadequate screening of women; however, almost 33% of failures to prevent invasive cervical cancer can be attributed to false-negative Pap smears, which can be attributed in part to poor quality of PTs.2 One of the major accomplishments of the original (1988) Bethesda System3 was to mandate the reporting of the specimen's adequacy for PTs.4 Although it was somewhat controversial at first,5 especially before well defined criteria of adequacy were agreed upon,6, 7 the reporting of specimen adequacy was embraced quickly by the majority of laboratories. Consecutive College of American Pathologists surveys indicated that the percentage of laboratories routinely reporting specimen adequacy for PTs increased from 35% in 1990 to 66% in 1991 and 85% in 1992.8, 9 The 2001 Bethesda System revised the criteria for adequacy assessment and, for the first time, also addressed the adequacy of liquid-based (LB) Pap specimens.10
The current American Society for Colposcopy and Cervical Pathology (ASCCP) management guidelines recommend a repeat PT within 2 to 4 months for women who have unsatisfactory PTs (UPTs).11-13 This recommendation is based largely on evidence derived from studies on conventional Pap smears. However, the significance of unsatisfactory LBPTs is uncertain, because the ‘obscuring’ factors (inflammation, blood, thick smearing) that typically are associated with unsatisfactory conventional Pap smears are less important in LBPTs. Furthermore, the evidence that there is an increase in abnormalities in the follow-up of women with unsatisfactory conventional Pap smears also is controversial, because the 2 studies that have attempted to address that issue using traditional and 2001 Bethesda adequacy criteria, respectively, reached different conclusions.14, 15
The objective of the current study was to asses the frequency, causes, and significance of unsatisfactory SurePath PTs, as defined by the 2001 Bethesda System, in a population that was screened predominantly by LBPTs. In addition, we compared the incidence of squamous abnormalities that occurred within a 2-year follow-up period in patients who had unsatisfactory SurePath PTs with the incidence of a matched group of patients who had negative SurePath PTs that were accessioned during the same period.
A search of our computerized cytopathology database (Mysis CoPath; Mysis Healthcare, Raleigh, NC) identified all PTs that were designated as unsatisfactory during the 48 months from January 1, 2003 to December 31, 2006. Conventional PTs were collected with spatula, endocervical brush, or a combination of the 2. The specimen was smeared on a glass slide that was labeled with the patient's name and medical record number and was fixed with spray fixative. SurePath specimens were collected with a Rovers Cervex Brush (Rovers Medical Devices B.V. Oss, the Netherlands) either alone or in combination with a Rovers EndoCervex-Brush. The tip(s) of the sampling device(s) was/were then placed in a BD SurePath collection vial, processed on a BD PrepStain Slide Processor (both from BD Diagnostics-TriPath, Burlington, NC), and held in alcohol for staining. ThinPrep PTs were collected using a broom-type or cytobrush/spatula cervical sampling device, which was rinsed in a ThinPrep vial that contained PreservCyt transport medium, and processed using a ThinPrep T2000 processor (all from Cytyc Corporation, Marlborough, Mass). Processed specimens were held in alcohol pending staining. Progressive Pap staining was performed on a Sakura DRS 601 automatic slide stainer (Sakura Finetek USA, Inc., Torrance, Calif) for all PTs.
These cases were evaluated and classified as unsatisfactory by 1 of 15 different cytotechnologists who were trained in interpreting both SurePath and ThinPrep PTs. An attempt was made in all cases to reprocess the LBPT and obtain another slide. The decision to reprocess was made by either the cytotechnologist or the pathologist who reviewed the case. Reprocessing was done according to the manufacturer's recommendations. According to the laboratory's ‘remake’ logs, 1776 LBPTs were reprocessed during this 4-year interval, and 1634 (92%) of those cases were reprocessed because of scant cellularity. Cases were diagnosed as unsatisfactory only if this second slide did not fulfill Bethesda 2001 adequacy criteria. According to institutional policy, cases that were classified as unsatisfactory by a cytotechnologist were reviewed by a second, senior cytotechnologist as a quality-control measure, and a final diagnosis of unsatisfactory was made only after the second cytotechnologist or a pathologist reviewed the case. Our institution uses less stringent criteria of adequacy in posthysterectomy vaginal PTs, especially in patients who also have received radiation and/or chemotherapy for gynecologic malignancies. The reason for the UPT designation was recorded. Clinical data were retrieved from archived, computerized PT records derived from physical or electronic requisition slips, including the woman's age, previous abnormal PTs (defined as PTs interpreted as atypical squamous cells of undetermined significance [ASC-US] or worse within the preceding year), physiologic state (pregnant, postpartum, menopausal), hormone therapy (hormone-replacement therapy, oral contraceptives, Depo-Provera), hysterectomy state, and previous genital tract malignancy with radiation and/or chemotherapy. The provider who collected the PT also was recorded along with the type of PT (conventional, SurePath, or ThinPrep) and whether the PT was obtained as a screening or as a diagnostic test. The designation of a PT as screening or diagnostic was made by the provider based on the Medicare rules,16 which stipulate that a diagnostic PT and related, medically necessary services are covered under Medicare Part B when ordered by a physician in women who have a personal history of cervical, uterine, or vaginal cancer that has been or currently is being treated; a previous abnormal PT; any abnormal finding of the vagina, cervix, uterus, ovaries, or adnexa; any significant complaint by the patient referable to the woman's reproductive system; or any signs or symptoms that, in the physician's judgment, reasonably may be related to a gynecologic disorder. Provider characteristics (ie, whether they were a physician, physicians' assistant [PA], nurse practitioner [NP], or certified nurse-midwife [CNM] and, for physicians, whether they were a resident or staff physician and the specialty in which they practiced) were obtained from the locoregional provider registries, from the Fairview System website, or from the clinics in which they practiced. All PTs that were accessioned with a number consecutive to the UPT accessioning number were retrieved. From this second group, a control group was formed of the PTs that remained after excluding those diagnosed as unsatisfactory or abnormal.
Next, the cytology and surgical pathology database (Mysis CoPath) was reviewed for any additional PTs that were performed up to 24 months after the index PT to determine the number of patients in who had squamous abnormalities found, including ASC-US, low-grade and high-grade squamous intraepithelial lesion (LSIL and HSIL, respectively), and invasive carcinoma, as well as glandular abnormalities (endocervical adenocarcinoma in situ, adenocarcinoma, or endometrial hyperplasia). Cervical biopsies and, in women who were status posthysterectomy, vaginal biopsies as well as endometrial biopsies and curettings that were obtained up to 24 months after the index PT also were retrieved and entered into a spreadsheet. All follow-up information was gathered up to March 31, 2008; therefore, 39 women who had UPTs and 42 women who had PTs with no evidence of intraepithelial lesion or malignancy (NILM) had less than 24 months of follow-up information (range, 16-23 months).
The cytology database also was queried for the number and type of PTs procured by each individual provider during this interval. Consecutive numbers were assigned for each provider, and the total number of PTs, the predominant type of PT methodology used, and the number of UPTs obtained by the provider were entered into a spreadsheet for further analysis.
All data were entered into an Excel spreadsheet (Microsoft, Redmond, Wash). Data were imported into SPSS software (version 14; SPSS, Inc, Chicago, Ill) for analysis. Analysis of UPTs and various characteristics were performed on the subset of clinicians that had performed at least 50 PTs. Graphic analysis of the total number of specimens by provider versus the unsatisfactory rate indicated a much higher and different pattern of unsatisfactory rates for this group. Appropriate contingency tables were constructed to compare various characteristics between the UPT group and the control group. These included providers' specialties, predominant PT method used, physiologic characteristics of the patient, and follow-up Pap and biopsy diagnoses. Analysis to determine the odds ratios (ORs) and 95% confidence intervals (CIs) of various characteristics in relation to the satisfactory status of the specimen was accomplished using logistic regression. The appropriateness of each analysis was assessed using the Hosmer-Lemeshow goodness-of-fit test. There was no indication that the analyses were not appropriate for the data. An assessment also was made to determine whether the number of total PTs collected correlated with a provider's unsatisfactory rate. All providers with any level of unsatisfactory specimens were included in the analysis. Because of the skewed distributions of both measures, data were converted to ranks, from lowest to highest, and plotted, as shown in Figure 1. The level of association was determined using a Spearman correlation. Then, 2 × 2 tables were constructed from the data on the UPT group and the comparison (control) group. A chi-square test or Fisher exact test (when the expected count of a cell was <5) was used to test for differences between groups, except for the differences between laboratory's cumulative 4-year PT diagnoses and the follow-up PT diagnoses of women with UPTs, for which a Mantel-Haenszel chi-square test (linear-by-linear association chi-square) was used. For all analyses, the level of significance was set at <.05. ORs are presented along with 95%CIs. ThinPrep and conventional Pap smears (CPS) were excluded from the analyses, except for providers' data analysis, in which all PTs were considered.
From January 1, 2003 to December 31, 2006, in total, 243,006 PTs (95.48% SurePath, 2.28% CPS, and 2.23% ThinPrep) were processed by our cytology laboratory, and 548 (0.23%) PTs were reported as unsatisfactory. During this interval, our laboratory, which serves mostly clinics that belong to the Fairview Health Services, a large, multihospital, regionally integrated, not-for-profit healthcare network that serves communities throughout Minnesota, diagnosed 5.3% of PTs as ASC-US, 0.5% as atypical squamous cells, cannot rule out HSIL (ASC-H), 2.1% as LSIL, and 0.4% as HSIL. The 15 individual cytotechnologists screened from 543 to 26,862 PTs during this interval and had UPT rates ranging from 0% to 0.46%; when the 4 cytotechnologists who screened <5000 PTs during this interval were excluded, the UPT rates were from 0.08% to 0.46%. These differences can be explained by the fact that cytotechnologists preferentially screened slides from different clinics with different patient populations and using different PT preparations.
Of the PTs that were classified as unsatisfactory, 106 cases were not processed, because no sample was collected, but an LBPT vial labeled with the patient's name was sent erroneously to the cytology laboratory. Such cases were identified by the cytopreparatory staff because of the absence of the sampling device tip in the SurePath vial and the presence of clear (rather than turbid) fixative fluid; the clinic was contacted and confirmed that no sample was collected. These cases were excluded from further analysis along with an additional 14 cases, which represented repeat UPTs of women who already were in the study, leaving 428 cases. All cases that were diagnosed as abnormal (ie, not as NILM) were excluded from the control group (28 ASC-US, 10 LSIL, 2 HSIL, and 4 unsatisfactory for evaluation). Because the main purpose of this study was to analyze the reasons and significance of unsatisfactory SurePath PTs, 26 CPSs and 124 ThinPrep PTs were excluded from the UPT (study) group, and 24 CPS and 101 ThinPrep PTs were excluded from the control group, which left 278 women in the UPT (study) group and 284 women in the control group. The clinical features of women in the UPT study and the control group are illustrated in Table 1. There were more diagnostic PTs (51 of 278 tests; 18.35%) in the UPT group than in the control group (33 of 184 tests; 11.6%; P = .03) (Table 2) Scant cellularity was the most common reason for the UPT diagnosis; and, in 19 cases, scant cellularity was combined with a second cause (Table 3).
|Unsatisfactory Group||Control Group|
|Age range (mean), y||16-88 (48)||15-83 (40)|
|Malignancy, radiation, chemotherapy||10||3.60||0||0|
|Oral contraceptive pill||16||5.76||36||12.68|
|Unsatisfactory Group (n=278)||Control Group (n=284)|
|Type of Pap Test||No.||%||No.||%|
|Obscuring foreign material||2||0.72|
|Obscuring endometrial cells||1||0.36|
|Scant squamous component||7||2.52|
During the study period, as a percent of all SurePath PTs that were processed, 0.16% (360 of 232,022 tests) were unsatisfactory. SurePath UPTs were statistically significantly more frequent in older women (age 48.4 ± 17.5 years vs 39.6 ± 15 years; P < .001) and were more likely to be from women who were menopausal (OR, 2.65; 95% CI, 1.7-4.2; P < .001) and status posthysterectomy or posthysterectomy with radiation and/or chemotherapy for malignancy (OR, 3.36; 95% CI, 2.0-5.5; P < .001). Patients in the UPT group less frequently were on contraceptives (OR, 0.40; 95% CI, 0.23-0.70; P < .001) or pregnant (OR, 0.44; 95%CI, 0.20-0.98; P = .04) compared with the control group.
The number of women who had a previous abnormal PTs in the UPT group was similar to that in the control group (31 of 278 women [11.2%] vs 27 of 284 women [9.5%]; P = .6). The majority of women who had UPTs (194 of 278 women; 69.8%) had cytologic and/or histologic follow-up. Repeated PTs were performed in 181 women in the UPT group (65.1%) versus 64 women in the control group (22.5%; P < .001) with an interval to follow-up PT of 3 months versus 12 months, respectively (P < .001). The UPT group was more likely to have abnormal follow-up PTs (14 of 278 women vs 4 of 284 women; OR, 4.0; 95% CI, 1.3-12.2; P = .01) than the control group (Table 4), and these women were more likely to undergo biopsy than women in the control group (29 women [10.4%] vs 6 women [2.1%]; OR, 5.3; 95% CI, 2.2-13.2; P < .001). The UPT group had a higher incidence of abnormal biopsies (cervical intraepithelial neoplasia 1 [CIN-1] or higher; 14 of 278 women [5%] vs 3 of 284 women [1%]; OR, 5.0; 95% CI, 1.4-17.5; P < .001) (Table 5). Among the abnormal biopsies, there were 7 significant lesions in women who had UPTs (CIN-2 in 1 woman, CIN-3 in 4 women, and endometrial hyperplasia in 2 women), but only 1 woman had CIN-2 in the control group.
|Unsatisfactory Group||Control Group|
|Follow- Up Pap Diagnosis||No.||%||No.||%|
|Follow-up Pap interval (mean), mo||0-24 (3)||0-24 (12)|
|No follow-up Pap test||97||34.89||220||77.46|
|Unsatisfactory Group||Control Group|
|Follow-Up Biopsy Diagnosis||No.||%||No.||%|
|Follow-up biopsy interval (median), mo||0-24 (5)||0-24 (2)|
|Endometrial hyperplasia with or without atypia||2||0.72||0||0|
The UPT group also had more severely abnormal follow-up PT results than the entire population of women screened by our laboratory during the same period (linear-by-linear association; P = .003). Excluding unsatisfactory cases, follow-up Pap diagnoses of LSIL (8 of 177 cases; 4.52%) and HSIL (1 of 177 cases; 0.56%) were higher in the UPT group than the 4-year laboratory rates of 2.10% for LSIL and 0.40% for HSIL.
The 243,006 PTs that were processed in our laboratory during the study interval were collected by 698 different healthcare providers, including 670 providers from clinics that used SurePath, 11 providers from a single obstetrics/gynecology clinic that used predominantly ThinPrep, and 17 providers from clinics that predominantly used CPS during the study period. The specialty of 220 of these providers, who collected 1.7% of all PTs during this period, could not be determined either because of common names that could not be assigned to a single provider or because the name of the provider was not found in any of the sources examined, most likely because he/she was not a local provider or was a resident. The remaining 98.3% of PTs were collected by 478 providers. Of these, 413 were physicians (MD or DO) and collected 81% of PTs, 18 were PAs and collected 5% of PTs, 31 were NPs and collected 11% of PTs, and 16 were CNMs and collected 3% of PTs. The physicians' specialties were family medicine (n = 195), obstetrics-gynecology (n = 83), internal medicine (n = 82), pediatrics (n = 14), and other (mostly surgical) specialties (n = 14). Obstetrician-gynecologists collected 37% of the PTs, family practitioners collected 35%, internists collected 8%, and pediatricians collected 0.5%. In addition, there were 25 residents who collected 1% of the PTs.
The numbers of PTs of all types collected and the unsatisfactory rates of providers who collected ≥50 PTs of any type during this 4-year interval are listed in Table 6. Although all providers who used ThinPrep and 77% of providers who used CPS had UPTs, only 30% of providers who used SurePath had any UPTs (P < .001). The percentages of providers who had at least 1 UPT also differed statistically according to their occupation and medical specialty (P < .001).
|No. of Pap Tests|
|Occupation||No. of Providers||Total||Mean||Median||Range||Any UPT, %||% UPT|
|Predominant preparation used|
When all providers were included in the analysis, regardless of the method of PT collection or the number of PTs performed, overall, the UPT rate was significantly lower for family practitioners compared with obstetricians (0.13% vs 0.25%; OR, 0.5; 95% CI, 0.4-0.6; P < .001). When only providers in clinics that used predominantly SurePath were compared, the obstetricians' UPT rate was not significantly different from that of nonobstetricians (0.12% vs 0.14%; OR, 0.85; 95% CI, 0.65-1.1; P = 0.2); however, NPs and CNMs had significantly lower UPT rates than physicians (0.08% vs 0.13%; OR, 0.64; 95% CI, 0.43-0.96; P = .03), and PAs had a significantly higher UPT rate than physicians (0.20% vs 0.13%; OR, 1.5; 95% CI, 1.5-2.35; P = .04). These results may indicate that physicians serve a patient population that has more of the factors associated with UPTs than NPs or that NPs may have a better technique based on more experience, because they have a higher mean PT collection rate than physicians (1070 PTs vs 694 PTs) (Table 4). There was an inverse relation between UPT rates and the number of PTs collected (Spearman ρ −0.71; P < .001) when only providers who had any UPT were analyzed (Fig. 1).
The major finding of this study, in addition to a low unsatisfactory rate in our predominantly SurePath LBPT-screened population, is that UPTs more commonly were followed by abnormal PT results and significant histologic abnormalities than satisfactory SurePath PTs that were interpreted as NILM. Low cellularity of the specimen was the main reason for classifying SurePath LBPTs as unsatisfactory. The relation between UPTs caused by obscuring inflammation, blood, or other elements commonly observed with CPSs can be explained by the finding that abnormal cells were ‘hidden’ or obscured by these factors. The relation between specimens with low cellularity and subsequent abnormalities, however, is more difficult to explain.
It has been demonstrated that specimen cellularity correlates with increased cytologic detection of squamous abnormalities on PTs.17 However, the minimum degree of cellularity required for a sample to be designated as adequate has been a controversial issue18 and has been set rather arbitrarily at 5000 cells for LBPTs by the 2001 Bethesda System based on some experimental support.19 However, questions related to cellularity, such as whether relatively hypocellular LBPT specimens (ie, specimens with between 5000 and 20,000 squamous cells) are entirely adequate and whether the same cellularity thresholds should be applied to different LBPTs (SurePath, ThinPrep, Monoprep)20, 21 still await answers. Some authors have suggested that a higher cellularity, with a threshold set at 10,00022 and even 15,00023 well observed squamous cells, is a better indicator of specimen adequacy. Cellularity of the LBPT sample depends mainly on 3 factors: the skill of the specimen taker, the preparation method, and the anatomic and histologic characteristics of the tissues sampled (cervix or vaginal vault in posthysterectomy women) and their physiologic or pathologic state. The influence of the provider on specimen adequacy has been well documented in previous studies performed on CPS,24, 25 although whether this also applies to LBPTs has not been addressed previously. Studies have indicated that there were differences in the quality of Pap smears procured by individual providers and that these were dependent on the specialty of the smear taker,26 the number of Pap smears taken annually,27 and the length of training28; however, it also was concluded that obtaining adequate smears appears to be as much an art as a science.29
Individual provider variability in LBPT adequacy rates appears to be lower than the variability observed with CPSs, most likely because of more standardized PT sampling devices and procurement methodology and preparation of slides by the laboratory rather than by each provider. A recent study from Wales demonstrated that the introduction of LBPTs (ThinPrep) resulted not only in very low inadequate rates in all participating laboratories but also in the reduction in variability in inadequacy rates between laboratories compared with conventional cytology.30
Improved PT quality by implementation of the Lean method resulted in a significant increase in newly detected CIN after a previous benign PT,31 demonstrating that the quality of the PT is important even in populations that are screened predominantly with LB preparations. Lean methods have been implemented widely in the industry and also recently have been applied to healthcare, especially to the laboratory environment, because clinical laboratory processes are the healthcare processes that most resemble manufacturing-type processes. Raab and collaborators31 have used the Perfecting Patient Care system, a type of Lean quality-improvement method that focuses on immediate problem solving and reorganization of processes and workflow, with the ultimate goals of improving quality, reducing inefficiencies, and decreasing waste and unnecessary costs.32 Those authors used a Lean redesign of clinicians' cervical cancer screening practices and a checklist that focused attention on every step in the PT procurement process and allowed the clinicians to correlate clinical findings with diagnostic findings in an extremely detailed manner. This resulted in improved PT sampling quality and a 15% proportional increase in the rate of CIN detection, possibly accounting for some of the squamous intraepithelial lesions that ordinarily are missed with cytologic screening for cervical cancer.31 Because 19% (106 of 548) of the PTs that were diagnosed as unsatisfactory in the current study were because the sample was never collected and the vial was sent to the laboratory mistakenly, it is likely that the application of similar quality-improvement methods would have a significant impact in reducing the UPT rate in our population.
It has been demonstrated that Pap smears taken by obstetrician-gynecologists are of better quality, as judged by the presence of a transformation zone component,26, 33 than those taken by family practitioners, although unsatisfactory rates did not vary. A very recent study, however, indicated that nongynecology providers had a 3-fold higher rate of unsatisfactory LBPTs (ThinPrep) compared with gynecologists.34 Our data show rather low variability in unsatisfactory rates among different categories of providers (MD, PA, NP, CNM) and medical specialties (family practitioners, internists, obstetrician-gynecologists, etc). When only SurePath users were compared, the UPT rate was not significantly different between obstetrician-gynecologists and nonobstetrician-gynecologists. We assume that the differences between our findings and those of Cole and collaborators34 are because of the LBPT method used, because the SurePath method is affected less by the presence of factors such as mucus, blood,35 and lubricant36 than the ThinPrep PT, and these factors, in turn, may be provider-dependent.
Anatomic and physiologic variables that may influence the squamous cellularity of the sample include the surgical absence of the cervix and previous loop electrosurgical excision or cone procedures,37 the phase of the cycle, the presence of atrophy, and the presence and extent of ectopy, which is accentuated during pregnancy.38 Although the presence of extensive ectopy may allow sampling of abundant columnar cells, it also may hinder the procurement of a satisfactory squamous cellularity. Pathologic processes that alter the presence and/or distribution of intercellular adhesion molecules in the squamous epithelium also affect squamous cellularity. Adhesion molecules, including integrins and the cadherin-catenin complex, typically are abundant in the basal and parabasal regions of the normal squamous epithelium of the cervix, but they become sparse with maturation of the epithelium toward the surface, thus allowing desquamation. It has been demonstrated that the persistence of abundant E-cadherin throughout the thickness of the squamous epithelium, as observed in squamous intraepithelial lesions39, 40 and cervical infections,41 correlates almost perfectly with the apparent ‘nonshedding’ of the dysplastic epithelium that results in false-negative PTs.39, 40 The retrospective nature of our study allowed us to compare only some of these factors between the UPT group and the control group. In agreement with previous studies, we observed that women in the UPT group tended to be older, menopausal, and status posthysterectomy.13 Rather surprisingly, we also observed that the postpartum state (but not pregnancy) was associated with UPTs, whereas use of hormone contraceptives and pregnancy were associated with a lower risk of UPTs.
Cellularity of the sample also is influenced by the method used to prepare the LBPT. ThinPrep PTs produce higher UPT rates than SurePath PTs in most studies, most likely because they are more vulnerable to the effects of excessive blood35 and mucus and the presence of lubricants.36 Because the presence of blood in PTs may be associated with pathology,42 specimens that are rendered unsatisfactory by this mechanism may harbor significant abnormalities, as also demonstrated by the increased number of abnormalities in ThinPrep PTs that are rendered satisfactory after reprocessing.43, 44 Our unsatisfactory rate is comparable to that reported in the literature when studies using the Bethesda 2001 definitions of adequacy are included. Published ThinPrep unsatisfactory rates, for example, vary from 1.1%,45 to 1.97%,47 to 5.47%44 with the national 50th percentile at 0.4% and the 95th percentile at 1.9%,47 whereas SurePath rates vary from 0.2%46 to 0.24%48 with the national 50th percentile at 0.2% and the 95th percentile at 1%.47
We did not attempt to analyze the factors that contributed to rendering specimens unsatisfactory because of rejection before processing. We did not include specimens that were rejected by the laboratory before they were accessioned, such as those with wrong or missing labels. However, we observed that the submission of vials despite not having collected a sample was a rather frequent occurrence. A recent comparative study of SurePath LBPTs and CPS also indicated an increase in ‘rejected’ specimens because of mislabeled/unlabeled samples, leading to an increase in direct costs despite a significant reduction in the proportion of unsatisfactory tests with the implementation of LBPTs.49 That study also indicated, like ours, that scant cellularity is the main reason for SurePath UPTs.49
In the current study, the incidence of biopsy-proven, significant abnormalities occurring within a 2-year follow-up period in patients who had SurePath UPTs was higher than in the control group with satisfactory SurePath PTs diagnosed as NILM. However, this result may have been influenced by the lower rate of follow-up Pap testing and biopsies in the control group. This difference in verification rates (70% in the UPT group and only 22.5% in the control group) is 1 of the limitations of our study. The different verification rates are because of the recommendation that women who have UPTs undergo a repeat PT 2 to 4 months after the UPT, and this follow-up PT falls well within the 2-year follow-up interval that we chose for the current study; however, for women who have a normal index PT, a follow-up PT typically is recommended after 3 years if no other risk factors are present. Therefore, a longer follow-up period extending beyond 3 years would be preferable in future studies to allow for more follow-up PTs (and possibly biopsies) to accumulate in the control group. If we compare only women who had PT follow-up in the UPT group and the control group, then the rate of abnormal follow-up PTs is similar (8.47% vs 7.81%, excluding follow-up UPTs); however, the severity of these abnormal PT results is higher in the UPT group than in the control group (4.52% LSIL and 0.56% HSIL vs 1.56% LSIL and 0% HSIL, excluding follow-up UPTs). Similarly, if only women who had follow-up biopsies are considered, then the 2 groups would gave a similar proportion of follow-up abnormalities (14 of 29 women [48%] in the UPT group vs 3 of 6 women [50%] in the control group). However, even accounting for different verification rates, women with UPTs had a higher likelihood of having an abnormal follow-up PT result. Because the probability of an abnormal PT result would depend on both the probability of having a follow-up test and the probability that it would be abnormal, we can multiply these probabilities. In women who had UPTs, the probability of having a follow-up PT was 181 in 278 (.65), and the probability that the follow-up PT would be abnormal was 15 in 177 (.085). Therefore, the probability that a woman with a UPT would be diagnosed later as abnormal on a follow-up PT was .65 × .085 = .052. For the control group, the probability of having a follow-up PT was 64 in 284 (.23), and the probability that the follow-up PT would be abnormal was 5 in 64 (.078). Therefore, the probability that a woman with a normal PT would be diagnosed later as abnormal on a follow-up PT was .23 × .078 = .018.
Our findings support the ASCCP management guidelines that patients who have a UPT, even if they have a negative history of gynecologic disease, require a repeat PT within 2 to 4 months.13 If the repeat PT also is unsatisfactory, then further evaluation, including colposcopy and/or biopsy, may be required. Previous studies on CPS reached contradictory conclusions. Using the original Bethesda adequacy criteria, Ransdell et al14 observed that unsatisfactory Pap smears were more likely to be from high-risk patients and more often had significant lesions on follow-up than satisfactory Pap smears. A similar conclusion was reached in a recent study in which women who had an unsatisfactory index Pap smear had a higher hazard rate for histologic CIN-2/CIN-3 or invasive squamous cell carcinoma, especially during the first 2 years of follow-up.50 However, Adams et al15 observed no differences in the incidence of squamous abnormalities detected on follow-up between patients who had unsatisfactory Pap smears because of inadequate squamous cellularity, as defined by the 2001 Bethesda System, and patients who had satisfactory and negative smears.
Another limitation of our study is that it was conducted retrospectively using data that were collected over the previous 48 months. The exact nature of the populations served by different providers and provider classes was not known. This made it difficult to explain the observed differences in performance between providers. Our use of ThinPrep LBPTs was low and derived mainly from a clinic that served predominantly obstetric patients, which made our results with ThinPrep difficult to generalize. Therefore, we excluded ThinPrep LBPTs from all comparisons other than provider comparisons. Further studies from institutions where ThinPrep LBPT use is more prevalent are needed to evaluate the significance of unsatisfactory LBPTs. Additional research into provider characteristics also may yield useful information. UPTs were more frequently from either posthysterectomy and/or post-therapy women than control group PTs despite the use of less stringent cellularity criteria of adequacy for these specimens in our institution. We did not review these slides to determine their cellularity and the possible variation in the levels of cellularity that individual pathologists/cytotechnologists considered unsatisfactory in such PTs. We acknowledge, however, that such variations may have affected our results.
Many cytopathologists,13 including those practicing at our institution, accept lower cellularity in vaginal specimens, especially when atrophy is present and the woman is at low risk, because such specimens may have lower cellularity and still may be satisfactory.13 Because no peer-reviewed data are available on what constitutes an adequate vaginal PT, further studies are needed to define the exact levels of cellularity that are acceptable for vaginal specimens.
The acceptance of lower levels of cellularity for vaginal PTs than those required for cervical PTs may explain in part the low UPT rate in our population. Another factor was the systematic use of reprocessing of LBPTs with low cellularity, a high percentage of which were rendered adequate. Whether these specimens, rendered adequate after reprocessing, have the same diagnostic value as specimens with initial adequate cellularity is an open question that we are trying to answer in a future study by comparing their respective subsequent abnormal PT rates.
In conclusion, we observed a low overall unsatisfactory rate in our population that was caused mainly by low cellularity. The cytologic and/or histologic follow-up of these women revealed more abnormalities than the follow-up of women who had satisfactory SurePath PTs diagnosed as NILM. Older age, menopause, posthysterectomy status, and/or prior malignancy with radiation therapy were patient-related factors that contributed to unsatisfactory tests. The lowest unsatisfactory rates were achieved by NPs and CNMs followed by physicians and PAs, but the differences were very small. To our knowledge, this study is the first to comprehensively analyze unsatisfactory SurePath LBPTs according to the Bethesda 2001 criteria, especially with regard to the factors that contribute to the unsatisfactory result. Our results support the recommendation of the ASCCP management guidelines to repeat the PTs or to perform other evaluations in women who have UPTs.
The authors made no disclosures.