Journalof urological method 29 (1990~267-278

267

Elsevier

VIRMET 01053

A comparison of cervical scrapes for

HPV typing by dot-blot hybridization
obtained by wood and plastic spatulas
M&ire A. Duggans, Masafumi Inoue, S.
Elizabeth McGregoP and Jill G. Nation
Departments of Pathology and 20bstetrics and Gynecology, Foothills Hospital,
The Tom Baker Cancer Centre and the Universityof Calgary and 3Department of
Epidemiology and Preventive Oncology, Alberta Cancer Board, Calgary, Alberta, Canada
(Accepted 11 May 1990)

Summ~y
A cross-over study was designed to determine whether the type of spatula used
to collect cervical cells influences the ability of. dot-blot hybridization to detect
HPV DNA. Fifty-nine patients had a cervical scrape with a wood spatula first
and a plastic spatula second: 60 were scraped in the inverse order. The order of
sampling did not affect the HPV DNA positivity rate, which was nearly similar for
both wood and plastic spatulas (30 and 32%, respectively). Wood spatulas collected
more cells and greater than 1 x 10s cells more often than plastic spatulas (P =
0.001 and 0.06, respectively). Non-purple (negative) dots were more frequent in
samples obtained by wood than by plastic spatulas (P = 0.001). The study showed
that cervical cell collection by wood spatulas is preferred as they harvest more cells,
thus optimizing the sensitivity of the hybridization method, and the spatulas are
also more economical. Although they yielded more non-purple dots, a reduction
in these dots by using plastic spatulas did not result in a significantly increased
HPV positivity rate.
Cross-over
scrape

study; Plastic versus wood spatulas; Dot-blot hybridization;

Cervical

Correspondence to: M.A. Duggan, Department of Pathology, Poothitls Hospital, 1403 - 29 Sheet N.W.,
Calgary, Alberta, Canada T2N 2T9.
016845 10/90/$03.50@ 1990 Elsevier Science Publishers B.V. (Biomedical Division)

268

Introduction
Mounting evidence strongly implicates specific types of the human papillomavirus (HPV) in the genesis of cervical intraepithelial neoplasia (CIN) and
invasive cancer (Koutsky et al., 1988). If HPV typing is a future consideration in
patient management, the cervical scrape, as it is already an established cytologic
screening procedure, represents the ideal test sample. A dot-blot hybridization
method using biotinylated probes to determine the HPV DNA present in cervical
scrapes was developed by Inoue et al. (1989). Using this technique, a minimum
of 1 x lo5 cervical cells is considered necessary to provide sufficient cellular
DNA to evoke a positive reaction. Purple dots are read as a positive reaction,
while white dots and non-purple dots, e.g. yellow/pale brown are interpreted as
negative.
The issue of the relative merits of cytologic specimens obtained by wood and
plastic spatulas with regard to increased cell collection and improved diagnostic
yield is unresolved (Stock et al., 1988; Rubio et al., 1980). With regard to the
cervical scrape for dot-blot hybridization, some concern was raised that the wood
spatula may harvest too few cervical cells to elicit a positive result, and also that
it may result in an increased number of non-purple dots, which may obscure a
weak positive reaction. Therefore, a cross-over study was designed to compare
the effect of the number of collected cervical cells, and frequency of non-purple
dots on the HPV positivity rate in cervical scrapes obtained by wood and plastic
spatulas.

Materials and Methods

A cohort of 119 consecutive patients referred to the colposcopy clinics of the
Tom Baker Cancer Centre with an abnormal cervical smear, or seen in follow
up for treatment of a prior CIN had a cervical scrape taken with an Ayre wood
spatula (American Hospital Supply, Edmonton, Alberta) for routine cytology.
The first 59 patients (Group A) had additional cervical scrapes taken for dot-blot
hybridization, first with an Ayre spatula and secondly with a plastic Cyto-spatula
(CanLab, Toronto, Ontario). The next 60 patients (Group B) had additional cervical scrapes for dot-blot hybridization carried out in the inverse order, i.e. plastic
spatula first, wood spatula second. All scrapes were circumferential samplings of
the squamo-columnar junction, and performed by a single gynecologist (JGN).
Colposcopic examination then took place. All patients had a colposcopically
directed cervical biopsy or endocervical curettage.
Pathology
Cervical scrapes for routine cytology were fixed with cytospray (CanLab,
Toronto, Ontario) and screened for abnormal cytological features which were
then evaluated by a single cytopatbologist
(MAD).

269

All specimens for histolo~ were fixed in 10% phosphate buffered formalin,
processed routinely, embedded in paraffin and stained with hematoxy~n and
eosin. Following evaluation of the cytological smears and histological sections, a
summary patient diagnosis, based on the highest degree of cervical abnormality
was reached.
Dot-blot hybridization

Details of the method are described elsewhere (Inoue et al., 1989). Briefly,
spatula tips bearing the scraped cervical cells for dot blot hybridization were
immersed in a 3.0 ml solution of Tris EDTA to which 1.0 ml of sodium
d~ecylsulfate was added to dissolve the ceils. Cervical samples were digested
overnight with proteinase K (0.25 mg/ml, Bethesda Research Laboratories (BRL),
Maryland, USA) at 56OC, followed by phenol and ~hlorofo~iso~yl~cohol
extractions. DNA was precipitated with ethanol, suspended in 80 ~1 of Tris
EDTA (10 mM Tris-HCl, pH 7.5, 1 mM EDTA), and stored at 4C.
The number of cells in the specimen was estimated by agarose gel electrophoresis. The electrophoresis was carried out in the presence of ethidium bromide dye
(0.5 pg/ml), applying 7.5 V/cm for 30-60 min. Using a l/80 volume of the DNA
sample, total cellular DNA was estimated by comparing the intensity of staining
with ethidium bromide to that of a human fibroblast DNA sample previously
quantitated by spectrophotometry, and run as a control. For this study, all samples were hybridized, regardless of whether the signal intensity was less than,
equal to, or greater than 0.6 pg of DNA (approximately 1 x lcr-6 cells).
The sample volume was reduced to 20 ~1 aliquots and directly spotted to the
ni~ocellulose membrane. The membranes were denatured and then air dried and
baked for 60 min in an 80C vacuum oven.
Cloned HPV 6, 11, 16 and 18 was obtained from Dr L. Gissman, Cancer Research Center, Heidelberg, F.R.G., and cloned HPV 33 from Dr G. Orth, Pasteur
Institute, Paris, France. Storage and culture of bacteria, plasmid amplification,
and purification of DNA was performed as described by Maniatis et al. (1982).
The purified DNA was subsequently labelled with biotin-1 I-dUTP by nick translation (Leary et al., 1983).
Membranes were incubated at 40C for at least 60 min in a prehybridization
solution (50% formamide, 4 x SSPE, 5 x Denhardts solution, 0.1% SDS and
0.1 m&ml carrier DNA). Hyb~dization was performed at 42OC for 16 h using a
biot~ylated probe concen~ation of 0.5 pg/ml and a reaction mixture containing
50% formamide, 4 x SSPE, 5 x Denhardts solution, 5% dextran sulfate, 0.1%
SDS and 0.1 mg/ml carrier DNA, and then overnight at -15OC. The membranes
were washed once for 15 minutes with 2 x SSC - 0.1% SDS at room temperature,
twice with 0.5 x SSC - 0.1% SDS for 30 min at -5OC, and then once with 2 x
SSC for 10 min. Thereafter, the membranes were incubated for 60 min at 63OC
with blocking solution (bovine serum albumin) and incubated with streptavidinalkaline phosphatase conjugate (BRL), followed by the chromogenic substrate
NBT/BCIP (BRL). Positive controls, composed of purified HPV DNA 6, 11, 16,

270

18 and 33 (100 pg target DNA and 200 ng carrier DNA) and a negative control
composed of 200 ng carrier DNA were run simultaneously with the test samples.
Cervical scrapes were read as positive for HPV DNA when clear, purple
dots could be distinguished from the background white color of the membrane.
White dots and dot colors other than purple, e.g. yellow or pale brown were read
as negative (Fig. 1). This method has a minimum detection level of l-10 pg
of viral DNA. or 1 x 105 HPV DNA molecules.

Statistics
The study used a cross-over design to control for any effect of the order in
which the smears were taken on the characteristics examined. Each patient served
as her own control. Paired t-tests and the Wilcoxon signed-rank test were used to
compare the mean number of cells harvested and the number of non-purple dots
by spatula type. McNemars chi-square test was used to compare the rates of
HPV DNA positivity by spatula type. Repeated measures ANOVA was used to
control for the effect of diagnosis and type of patient visit on the number of cells
collected and non-purple dots present. A significance level of 5% was accepted.

Fig. 1. HPV 16 positive control is dotted in Column A. Patient samples positive for HPV 16 are present
in Column C (#361 and 362). (Dot-blot hybridization of cervical DNA samples with biotinylated HPV
16 DNA probe).

271

All statistical analyses were performed using BMDP Statistical Software.

Results
Forty-six patients were first time attenders of the colposcopy clinic while 73
were seen in follow-up visits. Just over half of the patients had a condyloma
or CIN on cervical smear or biopsy (52%), while the remainder had benign
changes or no observed pathological abnormality (48%). In those scraped with
a wood spatula first and plastic spatula second (Group A), 53% had a condyloma/CIN diagnosis and 47% had negative/benign changes. In those scraped in
the inverse order (Group B), 43% had a condyloma/CIN diagnosis and 57% had
negative/benign changes.
The cervical scrape characteristics by order of scraping are summarized in
Table 1. The HPV DNA positivity rate was higher in group B, but the positivity
rates and the distribution of HPV DNA types within groups using different
spatulas did not differ significantly. In both groups, wood spatulas collected
more cells and were more likely to yield samples with more than 1 x 105,
cells. For either group, scrapes collected by wood spatulas had significantly more
non-purple dots and occurred more frequently than in those obtained by plastic
spatulas.
The characteristics of all cervical scrapes collected by wood and plastic spatulas are summarized in Table 2. HPV DNA detection and type distribution were
almost identical for both spatula types. Wood spatulas collected a significantly
greater mean number of cervical cells than plastic spatulas (P = 0.001). In addition, wood spatulas collected more than 1 x lo5 cells more frequently than
plastic spatulas: this difference was marginally statistically significant (P = 0.06).
Slightly more than half the scrapes collected by wood spatulas and with a condyloma/CIN diagnosis had greater than 1 x lo5 cells (53%): 48% had less than
1 x lo5 cells. The corresponding figures for scrapes collected by plastic spatulas were 47 and 66%, respectively. Several scrapes had more than one nonpurple dot; mean = 1.44 per specimen with wood spatulas and 1.26 with plastic
spatulas. Scrapes by wood spatulas had a significantly greater mean number of
non-purple dots per specimen (P = 0.000) and non-purple dots were present
more frequently (P = 0.000) than in those collected by plastic spatulas. Repeated
measures ANOVA indicated that more cells were collected and more non-purple
dots were present in scrapes collected by wood spatulas after controlling for the
effects of cervical diagnosis and type of visit, i.e., first time or follow-up.
The HPV DNA detection rates for samples collected by either spatula and
analyzed by diagnostic group were not significantly different, but were higher in
those with a condyloma/CIN diagnosis (Table 3). With either spatula, cervical
smears with more than 1 x lo5 cells had almost identical HPV DNA positivity
rates. However, in scrapes with less than 1 x lo5 cells those collected by plastic
spatulas were HPV DNA positive slightly more often than those collected by
wood spatulas (41 vs 35%, respectively). With either spatula and regardless of

10

HPV 16/18/33

13

< 1 x 10 cells

63

22

78

17

20
3

1.1
27

18

1.2
41

11

13
2

46

30

70

19

22
3

McNemars chi-square test; bPaired t-test; Wilcoxon signed rank test.

Mean number
Present

Non-purple dots

1.8
37

1.3
46

;? 1 x ld cells

Mean number (x 16)

Cells collected

12
2

P value

0.002
0.012

nsa

nsb

ns

nsa

2.2
40

14

1.3
46

17

25
8

67

23

77

28

42
13

Plastic first

HPV 6/l 1

Presence

HPV DNA

Group B
Plastic second

wood first

scrape characteristics by order of cervical scraping

Group A

TABLE

Cervical

2.9
54

10

1.5
50

18

24
6

90

17

83

30

40
10

Wood second

o.oo5c
O.OOoa

ns

O.OOlb

ns

nsa

P value

28

HPV 16/18/33

cells + diagnosis

Condyloma/CIN
Negative/Benign

x Id

2.4
91

11
12

77

48
52

53
47

19

81

24

30
7

aMcNemars chi-square test; bPaired t test; CWilcoxon signed rank test.

Mean number
Presence

Non-purple dots

CondylomaKIN
Negative/benign

51
45

23

< 1 x ld cells

c I x id cells + diagnosis

2 1

1.44
96

Mean number (x 16)

2 1 x 10 cells

Cells collected

36
8

1.6
67

21
11

41
46

32

1.26
87

28

38
10

Presence
HPV 6/l 1

HPV DNA

Plastic

n
%

wood

TABLE 2
Characteristics of cervical scrapes by spatula type

56

66
34

47
53

27

73

24

32
8

Yo

O.ooo
O.OfKF

Its (0.06)a

O.OOlb

nsa

nsa

P value

274

the diagnosis, scrapes with less than 1 x lo5 cells were HPV DNA positive more
frequently than those with more than 1 x lo5 cells.

Discussion
Except for the study of Peng et al. (1988) who showed no difference in the
detection of HPV 16 and 18 by filter in situ hybridization in smears obtained
by Cytobrush and cotton swab spatulas, the effect of the type of spatula used to
collect cervical cells for HPV typing has not been fully addressed. The relative
merits of cervical scrapes obtained by wood and plastic spatulas to elicit a positive
reaction for HPV DNA using dot-blot hybridization was examined in this study.
Possible variation in the number of cells collected due to different sampling
techniques was allayed by having a single gynecologist perform all the scrapings
(Voogis et al., 1988). A cross-over study was used so as to compare two groups
of patients with the same pathological diagnoses, and to control for the possibility
of a diminution in the number of exfoliable cervical cells containing typeable
HPV DNA due to a multiplicity of cervical scrapings. As the order of cervical
scraping using different spatulas had no effect on either the HPV DNA positivity
rate or distribution of HPV DNA types in the two groups, a decrease in cells
with typeable HPV DNA did not occur.
The study showed no significant difference in the HPV DNA positivity rate
or distribution of HPV DNA types in cervical scrapes obtained by wood and
plastic spatulas. Either spatula is therefore equally effective in collecting cervical
cells for HPV DNA typing. However, as wood spatulas are more economical,
they are logically preferable. Furthermore, wood spatulas, regardless of diagnosis
or type of colposcopy clinic visit harvested significantly more cells than plastic
spatulas (P = 0.001) and were superior in collecting scrapes with more than 1 x
lo5 cells: this difference was marginally statistically significant (P = 0.06). The
cell scrapes obtained by wood spatulas were therefore more likely to contain the
optimal number of cells for hybridization.
In previous studies using dot-blot hybridization and biotinylated HPV DNA
probes (Inoue et al., 1989; Duggan et al., 1990), it was noted that nitrocellulose
membranes with dotted DNA from scrapes obtained by wood spatulas frequently
had yellow/pale brown dots, regardless of the presence or absence of a positivepurple dot. These non-purple dots were interpreted as negative, but it was of
concern that they were possibly masking a weak positive dot. Moreover, the
cause of these dots was obscure. In the current study, non-purple dots were
statistically more frequent in smears obtained with wood spatulas (P = 0.000).
Their predominance in these scrapes suggests that an extrinsic factor(s), e.g. wood
chip contamination of the specimen by the action of the lytic and digestive steps
of the hybridization technique may be responsible for some of these dots. As they
were also seen, however, to a lesser extent in scrapes by plastic spatulas, other
factors, be they intrinsic, e.g., old or fresh blood contaminating the specimen, or
extrinsic factors inherent to the plastic spatula itself, probably contribute to the

+ diagnosis

aMcNemars chi-square test.

CondylomaKIN
Negative/benign

< I x 16 celk + diagnosis

CondylomaKIN
Negative/benign

2 I x ld

< 1 x Id ceils

Cells collected
2 1 x 10 cells

Condyloma/CIN
Negative/benign

Cervical diagnosis

s/11
3112

17151
1l/45

28196
8/23

45
25

33
24

29
35

36
25

10121
3/l 1

lb/41
9146

25187
13/32

26162
12157

22162
14/.57

HPV positive

HPV positive
%

Plastic

Wood

TABLE 3
Influence of diagnosis and cell number collected on HPV DNA positivity rates

48
27

39
20

29
41

42
21

ns
rIs8

P value

276

production of these non-purple dots. The excess of non-purple dots in scrapes

by wood spatulas did not affect interpretation of the membranes as a reduction
in these types of dots by using plastic spatulas did not result in a significantly
increased HPV DNA positivity rate. Therefore, it is recommended that cervical
scrapes for HPV typing continue to be taken with wood spatulas.
Dot-blot hybridization using biotinylated probes to detect the HPV DNA in
cervical scrapes requires a minimum of 1 x IO5 cells per specimen (Inoue et
al., 1989). This ensures at least 0.6 pg of cellular DNA, which optimizes the
methods ability to evoke a positive hybridization signal. A positive dot occurs
when the cervical scrape contains either a single cell with more than 1 x 105
HPV DNA molecules, or a number of cells, albeit with lesser amounts of HPV
DNA exceed this critical figure when cumulated.
In agreement with previous studies, cervical scrapes with a diagnosis of condyloma/CIN were HPV DNA positive more often than those with a benign/negative
diagnosis (Koutsky et al., 1988; Duggan et al., 1990). With either spatula, this
pattern continued regardless of whether the specimen had greater than or less
than 1 x lo5 cells. With both spatulas, but more so with the plastic spatulas, the
unexpected finding that specimens with less than 1 x 10 cells were HPV DNA
positive more often than those with more than 1 x lo5 cells can be explained by
the fact that 48 and 66% of scrapes with a condyloma/CIN diagnosis, obtained
by wood and plastic spatulas respectively had less than 1 x lo5 cells. This suggests that these samples had sufficient cumulative viral DNA to evoke a positive
reaction. In order to avoid missing such positive cases, representing 7% of this
study group it is recommended that all cervical samples be hybridized regardless
of cell number. However, the reporting of HPV DNA negative samples with less
than 1 x lo5 cells, representing 13% of this study group should include a caveat
stating that the specimen contains too few cells for optimal hybridization and
therefore should be repeated.
In summary, the study showed that cell scrapes collected by wood spatulas
are equally effective as those collected by plastic spatulas for detecting HPV
DNA by dot-blot hybridization. Wood spatulas collect significantly more cells
and greater than I x lo5 cells more often than plastic spatulas, thus optimizing the
sensitivity of the hybridization method. However, in order to avoid missing HPV
positive cases with less than 1 x lo5 cells, it is recommended that all specimens
be hybridized. Although scrapes by wood spatulas had significantly more nonpurple dots, they did not interfere with the interpretation of the membranes.
Consequently, as wood spatulas are also cheaper, these are preferred for this
hybridization method.

Acknowledgements
The authors wish to thank Drs L. Gissmann and G. Orth for generously
providing the cloned HPV DNA. The nursing staff at the Colposcopy Clinic of
the Tom Baker Cancer Centre are acknowledged for their valuable cooperation.

211

The authors also acknowledge Carol Gowzd, Kaorie Saito and Emi Inoue for
their technical assistance, and Cathy Hesch for typing the manuscript.

References
Duggan, M.A., moue, M., McGregor, SE., Gabos, S., Nation, J.G., Robertson, D.I. and Stuart, G.C.E.
(1990) Non-isotopic HPV DNA typing of cervical smears obtained at initial colposcopy exam.
Cancer (in press).
Inoue, M., Duggan, M.A., Robertson, D.I. and Chang-Poon V. (1989) Non-isotopic detection of HPV
DNA in cervical smears using dot-blot hybridization. J. Virol. Methods 26, 159-170.
Koutsky, L.A., Galloway, D. and Holmes, K.K. (1988) Epidemiology of genital human papillomavirus
infection. Epidemiol. Rev. 10, 122-163.
Leary, J.J., Brigati, D.J. and Ward D.C. (1983) Rapid and sensitive calorimetric method for visualizing
biotin-labelled DNA probes hybridized to DNA or RNA immobilized on nitrocellulose. Proc. Natl.
Acad. Sci. USA 80, 40454049.
Maniatis, T., Fritsch, E.F. and Sambrook, J. (1982) Molecular Cloning! A Laboratory Manual. Cold
Spring Harbor Laboratory, Cold Spring Harbor, New York.
Peng, H-Q., Roth, P., Caussy, D. and Rawls, W. (1988) Comparison of the cytobrush and cotton swabs
in sampling cervical cells for filter in situ hybridization detection of human papillomavirus types 16
and 18 DNA. Acta Cytol. 32, 311-313.
Rubio, C.A., Berglund, K., Kock, Y. and Zetterberg, A. (1980) Studies on the distribution of abnormal
cells in cytologic preparations: making the smear with a plastic spatula. Am. J. Obstet. Gynecol.
137, 843-846.
Stock, R.J., Thurmond, A.I. and Passmore, A. (1988) A comparison between the ACCU-Pap device and
the extended-tip wooden Ayre spatula for cervical cytology sampling. Acta Cytol. 32, 307-310.
Vooijs, G.P., Elias, A., Van der Graaf, Y. and Van de Berg, M.P. (1986) The influence of sample takers
on the cellular composition of cervical smears. Acta Cytol. 30, 251-257.