THE EFFECTIVENESS OF CYTOLOGICAL RESCREENING IN THE [611955]

1
THE EFFECTIVENESS OF CYTOLOGICAL RESCREENING IN THE
REDUCTION OF FALSE NEGATIVE/POSITIVE PAP REPORTS

Elena C. Cernescu1, Gabriela Anton2, Simona Ruta 2,3, Costin Cernescu 2,4
1Obstetric and Gynecology Clinic, Cantacuzino Hospital, Bucharest ; 2“Stefan S. Nicolau”
Institute of Virology, Bucharest , 3Carol Davila University of Medicine and Pharmacy ,
Bucharest 4Romanian Academy

*Corresponding author: Costin Cernescu, MD, PhD
“Stefan S. Nicolau” Institut e of Virology, Bucharest,
Romanian Academy
81, Amurgului Street; Bucharest, 051983
e-mail: [anonimizat]
tel: 0740.176.984

2
Background. Cytological investigation of the cervix has proven to be a valuable tool in
the early detection of cervical cancer; however the high incidence of false negative or
false positive smear reports is an important drawback.
Objectives. To investigate retrospectively the value of partial rescreening methods as
tools for improving the sensitivity and specificity of Pap test routine screening.
Methods. Out of a total 4664 cervical sample examined by Pap test, 20% were randomly
selected and rescreened with a more detailed examining protocol by the same cytologist;
in addition, target ed rescreening of all samples with severe lesions was carried out.
Results. During initial testing, 478 smears (10.24 %) show ed cytological abnormalities: in
5.79% ASC- US; 3.32 L-SIL and 1.14% H-SIL. At random rescreening a significant
decrease in the number of negative smears (83.05% v s. 85.9%, p=0.036) was recorded,
together with an increase (7.68% vs. 5.79%, p=0.043) in the number of smears classified
as ASC- US. No significant differences were recorded for L-SIL or H-SIL samples.
Retrospective target ed rescreening of all 208 samples initially diagnosed as L-SIL and H-
SIL revealed 42 false positive results and 12 false negative ones. Errors were linked to
suboptimal smear preparation: scant cellularity, material in clumps, paucity of abnormal
cells, pale dyskarosis, and small microbiopsy like aggregates.
Conclusion. Partial random rescreening or target ed rescreening enables a better
interpretation of suboptimal prepared smears. Target rescreening allows a correct
detection of even low percentages of atypical cells. Other confounding factors, such as
the lab workload and the regional disease prevalence, can exert an important effect on the
correct classification of cytological lesions. ( 262 words )
Key words : Pap test, false reports, screening, target screening

3
Introduction. Cytological investigation of the cervix has proven to be a valuable tool in
the early detection of cervical cancer. An important drawback of cervical cytology is the
high incidence of false negative or false positive smear reports issued by cytology
laboratories [1]. Although the screening false negative rate is believed to be lower for
intraepithelial neoplasia than for invasive carcinoma, in some laboratories it may be as
high as 20% [2]. Traditional methods of internal quality control, such as random
rescreening or targeted rescreening of symptomatic cases can only make a limited
contribution to decrease the number of false negative reports [3].
The aim of this study was to investigate retrospectively the accuracy of the initial
classification of cytological lesions; both by partial random rescreening and by targeted
rescreening of smears with severe lesions and to examine whether there were particular
smear patterns that cause errors during the conventional screening.
Screening procedure and patients.
Cervical smears collected from all women (mean age 29.23 ± 0.7 years; limits: 19-53
years) attending an Obstetric and Gynecology Clinic in Bucharest for routine
gynecological exams during 2009-2010, except for the ones with previous abnormal Pap
tests, were screened for cytological abnormalities according to Bethesda protocol [4]. The
results were classified as: negative for intraepithelial lesions (NEG), atrophic lesions,
atypical squamous cells of undetermined significance (ASC-US), low-grade squamous
intraepithelial lesions (L-SIL), and high-grade squamous intraepithelial (H-SIL). For
routine screening (RS) the examination was done using the 20x objective, by step
screening the smear divided in two diagonal parts; about one microscope field being

4
examined at each step. This method allows the inspection of about 25% of the area under
the coverslip (fig 1); the screening procedure takes 2 to 3 minutes.
For rescreening, 20 % of the total number of samples were randomly selected and
reexamined by the same cytologist, with a more detailed protocol, without knowledge of
the previous results. The area under the coverslip was split in three sectors, those where
the cellular material was concentrated, six diagonals were examined with the 40x
objective; each step corresponding to one microscope field divided by two. By this
method, about 80% of the area under the coverslip is examined and the procedure takes
10 to 12 minutes.
The results of standard screening and rescreening were compared and if there was
consistent discrepancy between the two, the respective slides were reviewed by an
independent cytologist, and a final result was issued.
Results.
The laboratory workload. The working load of the cytopathology laboratory between
October 2009 and October 2010 is presented in fig 2. In total, 4664 cervical smears were
examined with the routine screening protocol (RS) by one cytologist, with a monthly
average of 388.6 samples (maximum: 540; minimum: 266 samples). The number of
slides examined under the routine screening protocol rarely exceeded 30 samples per day,
on average, the maximum number was 28.3 samples each Wednesday, and the minimum
was 21.6 samples each Friday.
The rate of abnormal smears in each month is shown in fig. 2. There were: 179 (3.84%)
atrophic samples; 270 (5.79%) samples classified as ASC-US; 155 (3.32%) samples
classified as L-SIL and 53 (1.13%) samples showing severe abnormality, fulfilling the H-

5
SIL diagnostic criteria. There was no correlation between the frequency of samples with
abnormal results and the lab’s monthly working load.
As an internal control, for quality improvement purposes, around 20% of the total
number of samples (820/4223) were randomly chosen and rescreened (RR-20) with the
protocol described in the method section.
Throughout this paper 'abnormal smear' is defined as one showing at least borderline
change or any grade of dyskaryosis; false negative refers to false normal, and false
positive refers to false abnormal or with more severe results at routine screening than
those recorded at rescreening.
Table 1 shows the detection rate of Bethesda cytological categories after routine
screening and after random rescreening of around 20% of samples. Important differences
at rescreening were registered in the proportion of samples initially diagnosed as negative
for cytological lesions (83.05% vs. 85.9%, p=0.036, RR=1.036, 95%CI=1.001-1.071), or
as ASC- US (7.68% vs. 5.79%, p=0.043, RR=0,9505, 95%CI0.9013-1.002 ).
Targeted rescreening of samples with severe lesions. As the differences between
samples classified as L-SIL or H-SIL at RS and RR-20 did not reach statististical
significance (table 1), a targeted rescreening (TR) of all samples initially diagnosed with
severe lesions was performed by an independent cytologist.
Out of the total 208 samples initially diagnosed as L-SIL or H-SIL, 42 (20.1%) were
judged as false positive, and downgraded to negative (14 samples) or ASC-US (28
samples) (table 2). Two samples diagnosed as H-SIL were downgraded to L-SIL.
In most cases that were finally considered consistent with lower grade abnormality, the
discrepancies were attributed either to the scarcity of abnormal cells or to the presence of

6
cells obscured by inflammatory context. Twelve smears initially diagnosed as L-SIL or
HSIL were re were finally classifi ed as H-SIL after targeted rescreening. Overall, the rate
of false positive results was 0.9% and that of false negative results was 0.21, out of the
total 4664 initially screened samples.
Analysis of discordant results for severe abnormalities at routine vs. targeted
screening. In order to clarify the reasons that determined discordant interpretations, all
observations concerning inadequate sampling or other confounding factors for
moderate/severe dysplasia were recorded. Cell configurations associated with
controversial results were analyzed to determine whether there are particular smear
patterns habitually misinterpreted on routine screening. Table 3 presents the most
frequent patterns for all cytological categories, with material distributed in clumps, scarce
cellularity and changes obscured by hematies or inflammatory cells being constantly
noted. A high frequency of metaphase cells showing enlarged nuclei with irregular
borders, but no hyperchromasia as well as the presence of immature metaplasic cells with
borderline nuclear change were the main criteria used to differentiate low and high grade
squamous intraepithelial lesions. Koilocytes, as well as bi/multi nucleated cells were
present in less than half of abnormal smears; their presence cannot indicate with certainty
a specific cytological result. No statistically significant correlation was found between the
frequency of koilocytes and the ascertainment of samples to low or high grade squamous
intraepithelial lesions.
Discussion.
In order to optimize the clinical value of the Pap test, it is critical to increase the
laboratory performances both in sample collection and in the interpretation of the results.

7
Cytology screening has some significant obstacles related to the shortage of highly
qualified, experienced cytologists/cytotechnicians; the subjective component in the
cytological diagnosis; the monotonous cytologist activity especially in highly screened
populations, with low prevalence of lesions. Cytopathology laboratories use several
quality control methods to decrease the frequency of Pap test false-negative (FN) or false
positive (FP) results. In the United States, the 1988 Clinical Laboratory Improvement
Amendments of 1988 (CLIA ’88) regulations require that a minimum of 10% randomly
selected Pap tests classified as negative for intraepithelial lesion or malignancy (NILM)
in the routine screening (RS) process are rescreened (R-10% method) before case sign-
out and verification [17]. An alternative quality control method, used predominantly in
Europe, is rapid pre-screening (RPS) of all smears, for 30 to 120 seconds before RS [16],
in which the cytotechnologist makes an interpretation of normal or atypical without any
markings on the slide. Tavares et al. [18] showed that RPS detected significantly more
FN Pap tests and increased RS sensitivity (71.3% to 92.2%) compared with R-10%, a
finding confirmed also by other studies [19, 20].
Partial random rescreening or full rescreening of samples with severe lesions
(higher than L-SIL) are accessible internal quality control methods that allow the
detection of errors due to sampling errors or misinterpretation and can reduce false
negative or false positive report rates from 2.5% to less than 0.9% [7]. In our study, the
percentage of abnormal smears that were overcalled (suspected of being more severe on
standard screening) was 0.94%, four times higher than the percentage of under-called
results. Concerns related to the poor sensitivity of cytology-based screening, can often
lead to a more frequent calling of ASC-US, but at the expense of a lower specificity. Our

8
data show that targeted rescreening involves a more vigilant examination that is likely to
accurately detect rare events and to perceive errors secondary to gaps in concentration. A
number of factors may account for the divergent sets of data recorded at rescreening.
First, both methods used to improv e the sensibility of screening (targeted rescreening of
previously abnormal samples or random rescreening of a limited proportion of all tested
smears) allow the re-examination of only a small proportion of negative smears, thus
having a minor contribution in limiting the number of false negative reports. The smears
that were erroneously upgraded to moderate or severe dysplasia were generally of poor
quality and showed a few dyskaryotic cells, pale dyskaryosis or microbiopsy.
Second, one of the greatest clinical anxieties related to cervical screening is the
management of low grade cytological abnormalities, namely borderline nuclear changes
and mild dyskaryosis [13]. These abnormalities are of concern, because in a small, but
nevertheless important, number of cases, they can reflect the presence of persistent HPV
infections with high risk genotypes. With the establishment of the links between HPV
infection and cervical cancer [10, 11], new directions have been proposed for
rescreening, such substituting cervical cytology screening with HPV DNA testing or
combining both methods (co-testing). The superiority of HPV testing in primary
screening compared to cytology has not been demonstrated in several randomized clinical
trials [14, 16]. Although HPV screening was proved to be more sensitive than cytology in
detecting CIN3+/CIN2+, it was also less specific. Evidence showed that the expected
harms (e.g. high rate of (false) positive test in normal cytological samples from young
women, followed by promotion of unnecessary procedures like colposcopy, and possible
problems with future pregnancies) outweighed the potential benefits [14, 15]. In countries

9
with organized cervical screening programs it is considered unethical to have a high
proportion of false positive smears, which result in over-treatment and in an unnecessary
psychological burden for patients. Several clinical trials, involving women aged 30 years
and older, showed mixed results for co-testing also compared to cytology alone. (ref). As
HPV infection alone does not necessarily conduct to cervical cancer (infection rates are
approximately 1,000 times higher than the incidence of cervical cancer and at least 90%
of HPV infections resolve spontaneously [12]), there is a continuing need to validate
HPV tests and to establish which ones are acceptably reproducible, accurate, and cost-
benefit balanced. The effect of other factors, such as the cytotechnologist workload and
the regional disease prevalence, must also be taken in consideration in order to decide the
best strategy that will distinguish true precursor lesions from those that have no clinical
relevance.

References
1. International Agency for Research on Cancer. Cervix Cancer Screening. IARC
Handbook of Cancer Prevention, vol. 10. Lyon, France: IARC Pr; 2005.
2. Arbyn M, Antilla A, Jordan J, Ronco G, Schenck U, Segnan N, et al .
European Guidelines for Quality Assurance in Cervical Cancer Screening. 2nd ed.
Luxembourg: Office for Official Publications of the European Communities;
2008.
3. Van der Graaf Y, Vooijs GP, Gaillard HL, Go DM . Screening errors in
cervical cytology screening. Acta Cytol. 1987; 31: 434-8.

10
4. Solomon D, Davey D, Kurman R, Moriarty A, O’Con nor D, Prey M, Raab S,
Sherman; M, Wilbur D, Wright T Jr, Young N: Forum Group Members
Bethesda 2001 Workshop. The 2001 Bethesda System: terminology for reporting
results of cervical cytology. JAMA 2002, 287(16) :2114 –2119.
5. Arbyn M, Schenck U. Detection of false negative Pap smears by rapid
reviewing: a meta-analysis. Acta Cytol . 2000; 44: 949-57.
6. Arbyn M, Schenck U, Ellison E, Hanselaar A. Metaanalysis of the accuracy of
rapid prescreening relative to screening of Pap smears. Cancer 2003; 99: 9-16.
7. Krieger P, Naryshkin S . Random rescreening of cytologic smears. A practical
and effective component of quality assurance programs in both large and small
cytology laboratories. Acta Cytol. 1994; 38: 291-8.
8. Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The
development of QUADAS: a tool for the quality assessment of studies of
diagnostic accuracy included in systematic reviews. BMC Med Res Methodol .
2003;3:25.
9. Saraiya M, Berkowitz Z, Yabroff KR, Wideroff L, Kobrin S, Benard V.
Cervical cancer screening with both human papillomavirus and Papanicolaou
testing vs. Papanicolaou testing alone: what screening intervals are physicians
recommending? Arch Intern Med . 2010; 170:977-85.
10. Kinney W, Stoler MH, Castle PE. Special commentary: patient safety and the
next generation of HPV DNA tests. Am J Clin Pathol. 2010; 134:193-9. [PMID:
20660320]

11
11. Mayrand MH, Duarte-Franco E, Rodrigues I, Walter SD, Hanley J,
Ferenczy A, et al; Canadian Cervical Cancer Screening Trial Study Group.
Human papillomavirus DNA versus Papanicolaou screening tests. N Engl J Med.
2007; 357:1579-88.
12. U.S. Preventive Services Task Force Liquid-Based Cytology and Human
Papillomavirus Testing to Screen for Cervical Cancer (October 2011) accesed at
http://www.uspreventiveservicestaskforce.org/uspstf11/cervcancer/cervcancerupd
.htm
13. Anttila A, Pokhrel A, Kotaniemi-Talonen L, Hakama M, Malila N, Nieminen
P. Cervical cancer patterns with automation-assisted and conventional cytological
screening: a randomized study. Int J Cancer. 2011; 128: 1204-12.
14. Kitchener HC, Fletcher I, Roberts C, Wheeler P, Almonte M, Maguire P . The
psychosocial impact of human papillomavirus testing in primary cervical
screening-a study within a randomized trial. Int J Gynecol Cancer. 2008; 18:743-8.
15. McCaffery KJ, Irwig L, Turner R, Chan SF, Macaskill P, Lewicka M, et al .
Psychosocial outcomes of three triage methods for the management of borderline
abnormal cervical smears: an open randomised trial. BMJ . 2010; 340:b4491
16. Franceschi S, Denny L, Irwin KL, Jeronimo J, Lopalco PL, Monsonego J, et
al. Eurogin 2010 roadmap on cervical cancer prevention. Int J Cancer.
2011; 128:2765-74.
17. Department of Health and Human Services, Health Care Financing
Administration . Clinical laboratory improvement amendments of 1988: final
rule. Fed Regist. 1992; 57:7146.

12
18. Tabbara SO, Sidawy MK . Evaluation of the 10% rescreen of negative
gynecologic smears as a quality assurance measure. Diagn Cytopathol.
1996; 14:84–86.
19. Brooke D, Dudding N, Sutton J. Rapid (partial) prescreening of cervical smears:
the quality control method of choice? Cytopathology. 2002; 13:191–199.
20. Lee BC, Lam SY, Walker T . Comparison of false negative rates between 100%
rapid review and 10% random full rescreening as internal quality control methods
in cervical cytology screening. Acta Cytol. 2009; 53:271–276.

13
Fig 1 . The diagram of step screening and rescreening procedure

14
Fig.2. The monthly workload of the cytopathology laboratory and the distribution of
samples with abnormal results according to the main Bethesda categories
0100200300400500600700Total number of screened samplesMonthly workload in the cytopathology lab
All samples ASC -US L-SIL H-SIL

15
Table 1. Detection rates for cytological abnormalities by routine step screening
versus random 20% rescreening
Cytological results Routine step screening Random
rescreening(RR- 20) p*
Number % Number %
Negative 4007 85.9 681 83.05 0.0363
Atrophic 179 3.84 38 4.63 0.3263
ASC- US 270 5.79 63 7.68 0.0439
L-SIL 155 3.32 25 3.05 0.7637
H-SIL 53 1.14 13 1.58 0.3608
Total number of
slides 4664 820
*p<0.05 is considered statistically significant
a) Comparison between the results of 100% routine step screening (RS) and
20% random -rescreening (RR-20).

16

Table 2. Detection rates of false positive or false negative results after targeted
rescreening

Initial
diagnostic No of
rescreened
samples Diagnostic after target rescreening

Negative ASC- US L-SIL H-SIL
L-SIL 155 12 24 – 10
H-SIL 53 2 4 2 –
Total 208 42
12

17

Table 3. Detection rates for cytological abnormalities after target rescreening in
samples previously diagnosed as low or high squamous intraepithelial lesions.

Cytological profiles
Proportion of cytological changes recorded at targeted
rescreening
Negative:
N=14 (%) ASCUS:
N=28 (%) L-SIL:
N=119 (%) H-SIL:
N=47 (%)
Material distributed in
clumps 8 (57.1) 8 (28.6) 12 (10.1) 3 (6.4)
Scanty cellularity – 10 (35.7) 25 (21) 5 (10.6)
Changes obscured by
hematies or polymorphic
cells 6 (42.9) 4 (14.3) 35 (29.4) 2 (4.2)
Metaplasic cells – 2 (7.2) 17 (14.3) 37 (78.7)
Koilocytes – 2 (7.2) 43 (36.1) 20 (42.6)
Bi/multi nucleated cells – 2 (7.2) 35 (29.4) 19 (40.4)
Immature
metaplasic cells with
borderline nuclear
change – – 2 (1.7) 37 (78.72)