|
|
|
|
Leopold G. Koss, MD
|
|
|
The
complex detection system leading to the discovery and treatment of
precancerous lesions and early cancer of the uterine cervix is described
in detail and discussed. By far the most difficult and underestimated
component of this system is the screening and interpretation of cervical
(Papanicolaou) smears. Cytologic case finding may fail because of
inadequate samples, insufficient time devoted to screening, or human
fatigue. Other weak points of the system, such as an inadequate clinical
component, inadequate patient compliance, poor reproducibility of
diagnoses, and ineffective aftercare, are also described. For example,
obtaining a second smear to confirm or refute a diagnosis of cellular
alypia is often a misleading practice. Although this cancer detection
system has been shown to be effective in reducing the rate of morbidity
and mortality from invasive cervical cancer in appropriately screened
populations, there is no evidence that the Papanicolaou test has
succeeded anywhere in complete eradication of this theoretically
preventable disease. It is important to inform the public about the
potential failures of the system and the reasons for them. (JAMA
1989;261:737-743)
THE current flurry of interest in the Papanicolaou (Pap) test in the lay
press (Idyll Street Journal, Nov
2, 1987, p 1; Vidll Street
Journal, Dee 29, 1987, p 17; Newsweek'
Jan 25, 1988, p 54; Washingto1~
Insider's Focus, July 24, 1987, p 1) has its roots in the historical
development of this important and little‑understood laboratory
procedure. In or about the year 1924, George N. Papanicolaou, MD, PhD,'
an investigator interested in the endocrinology of the menstrual cycle,
made an incidental observation that carver cells derived from the
uterine cervix may be observed in human vaginal smears. He presented
this observation in May 1928, apparently unaware that a Rumanian
pathologist, Aureli
From the Department of Pathology, Montefiore Medir~al Center, Albert
Einstein College of Medicine, Bronx,
Reprint requests to Department of Pathology, Montef~oreMedical Center,
111 E 210th St, Bronx, NY 104672490 (Dr Koss) HA
February 3, 1989𦅛ol
261, No. 5 Babes,23 had introduced cytologic sampling of the uterine cervix for the diagnosis of cancer at least two years earlier and had published a detailed account of it in April 1928.
In 1939 Papanicolaou entered into an association with the Cornell
gynecologist Herbert Traut, MD.4 In the vaginal pool smears provided by
Traut, Papanicolaou identified cancer cells in a number of patients with
malignant tumors of the uterine cervix and endometrium, some that had
not been suspected clinically. The early results were presented in
1941,5followedbyabookin 1943.6The observations were soon confirmed by a
number of investigators from other institutions.7-" In 1947 a
Canadian gynecologist, J. Ernest Ayre, MD, documented that a sample
obtained directly from the uterine cervix by a wooden spatula was more
efficient and easier to examine than a vaginal smear. 12 Shortly after
the introduction of the test it was noted that cancerous changes still
confined to the epithelium of the uterine cervix (carcinoma in situ)
could be identified in the cytologic samples. 13 ]5 On the assumption
that treatment of these precancerous lesions would prevent invasive
cancer of the uterine cervix, the test was hailed as the ultimate tool
in cancer detection and prevention. Since the name Papanicolaou was too
long, the term Pap test was
coined and applied to the screening procedure that entered slowly into
the mainstream of laboratory testing.
The efficiency of the cervical smear has never been tested in a
prospective blinded study. There has been no objective statistical
analysis of the optimal performance of this procedure. The test has been
considered by most third‑party payers as equivalent to other
automated or semiautomated laboratory procedures, although it is a
uniquely labor‑intensive complex process, the outcome of which
depends entirely on human judgment and is not machine generated. Until
recently the accuracy of the test has been rarely questioned. This issue
of paramount importance to women has received scanty attention from
women's rights advocates in spite of concerns that have been repeatedly
expressed in scientific publications. ]6 19
I will attempt to describe the components and the achievements and
failures of the cervix cancer detection system, of which the Pap test is
a critical component, but not the only important one. CERVIX
CANCER DETECTION SYSTEM
Detection of precancerous states and small, curable, occult invasive
carcinomas of the uterine cervix is based on a complex system of
clinical and laboratory procedures. All the links in this chain,
described below, must function well for the system to perform optimally. Obtaining
the Sample
The cervical sample should contain cells from the squamous epithelium of
the vaginal portion of the cervix, from the squamocolumnar junction
(also known as the transformation zone), and from the endocervical
epithelium. Ideally, a sampling instrument should be selected that is
most appropriate for the anatomy of the individual patient. If a
commercial, mass‑produced instrument is used, it must be rotated
around the uterine cervix with considerable pressure, with the tip of
the instrument placed in the endocervical canal. An additional
endocervical sample may assist in the discovery of endocervical lesions.
In perimenopausal or postmenopausal women, a sample of the vaginal pool
cells should be obtained in addition to the cervical smear. This may
serve to identify cancer cells from the endometrium, tubes, and ovaries;
these cancers become increasingly common in women past the age of 50
years.
It is generally assumed that obtaining a cervical smear is an easily
executed, clinically simple procedure. This is not correct. To obtain a
cell sample from the uterine cervix the organ must be visualized with a
vaginal speculum introduced without lubricant (to avoid contamination of
the cell sample with foreign material). This is not an easy task in the
absence of training, nor is it necessarily painless to the patient if
performed by a medical person without suitable skills, such as
inexperienced house staff. In my experience, a well-trained paramedical
person usually performs better than most untrained physicians, who may
consider taking the smear a boring and unimportant function.
Obtaining the appropriate sample is only a part of the procedure;
preparation of the smear requires that both sides of the sampling
instrument be carefully and rapidly pressed to a clean glass slide;
otherwise, much of the sticky material remains attached to the
instrument. To prevent air‑drying artifacts, the smear must be
fixed rapidly, either by immersing it into a fixative, such as 70%
alcohol, for 15 minutes or by spraying the surface with one of the
commonly available fixatives held at an optimal distance of about 25 cm
between the nozzle of the spray bottle and the slide. All of these
procedures require dexterity and experience and should optimally be
performed by a trained assistant. Needless to say, appropriate
identification of all smears is essential.
The laboratory must also be provided with a summary of essential
clinical data, such as age, obstetric and gynecologic history, and a
description of clinical findings. Ideally, the assessment of risk
factors should also be included, although this is not customary today. Processing
the Sample
In the laboratory the smears must be logged into a recording system and
given an identifying number. Smears treated with one of the spray
fixatives must be washed before staining. The staining procedure with
the commonly used Pap stain requires that the solutions be either
freshly prepared on a daily basis or at least filtered to avoid the
contamination of smears by "floating" cells. The stained
smears should be protected by a glass coverslip; protecting smears with
liquid plastic may compromise the study of cytologic detail. After
drying, the slides, accompanied by a form containing the summary of the
clinical data, are forwarded to the screening laboratory. A summary of
the results of previous smears and/or biopsies should accompany the
smear. Screening
the Smears
The purpose of screening Pap smears is to identify, among the multitude
of cells present, those cells that may reflect abnormal cancerous or
precancerous epithelial changes. The smears may also disclose other
important findings, such as infections or infestations. All abnormal
smears should be reviewed by a pathologist for a final diagnosis and
recommendation to the clinician.
The task of screening is delegated to trained cytotechnologists who
attend specialized schools for one or two years and should pass a
qualifying examination given by the American Society of Clinical
Pathologists. Any person who has not screened a large number of
cervicovaginal smears has a limited notion of the physical and mental
effort the procedure requires. To do it well, every cell in the smear
must be viewed, and any abnormality must be recorded on the slide by a
system of markings with ink. Even for a very‑well‑trained
and talented cytotechnologist this is a timeconsuming task. If one
considers that each smear contains from 50 000 to 300 000 cells, the
number of observations required is stupendous. A careful screening of a
Pap smear requires at least five minutes per slide, rarely less, and a
difficult case sometimes requires considerably more time. Thus, even
under optimal circumstances and assuming no fatigue, no more than 12
smears can be screened in an hour, or, at most, 90 smears in a
7.5‑hour day or 450 smears in a 37.5‑hour week. Assuming a
48-week year (considering vacations and holidays), a single
cytotechnologist could screen a maximum of 21 600 smears annually. In
reality, adequate screening of 90 smears a day is a target only a few
technologists can attain. The fatigue is such that after an hour of
uninterrupted screening, a break of ten minutes is needed. Review of the
clinical data, sometimes crucial to interpretation, and the logging of
the preliminary diagnosis add still further to the time required to
handle each case. It is also important to consider that quite frequently
two smears per patient are submitted. Thus, in practice, screening of 50
gynecologic cases per day is a reasonable limit, resulting in 250 smears
in a week and 12 000 smears annually. This limit, which admittedly is
anecdotal and not based on a controlled study, was: proposed by the
American Society of Cytology.
This target is observed in many lab- oratories dedicated to the
excellence of the procedure, but, according to the reports in the lay
press (Wall Street Jo~`rnal, Nov. 22, 1987, p 1), it is frequently
disregarded. The impact of rapid screening of a larger number of smears
on the accuracy of the procedure is unknown. I am aware of only one
study,? published only as a meeting abstract, in which a superbly
talented cytotech:3 nologist, the late Andrew Ricci, work‑out ing
with Thomas R. Simon, MD, undertook to screen 4184 cases, each
consisting of a cervical and vaginal smear; the screening time was
limited to no more than two minutes per smear, or four minutes per case.
The smears were classified as inadequate, normal, or requiring further
attention. Each slide was subsequently rescreened in routine fashion. Of
the 4046 adequate cases, 3484 (86.1%) were considered
"normal," and 562 (13.9%) were considered to require further
attention. On rescreening of the "normal" cases, there were
two endometrial carcinomas and three cases of significant abnormalities
of cervical squamous cells missed on the first screening. Among the 562
cases tagged for further study, there were 17 cases of preneoplastic
lesions of the uterine cervix (3.0%). This study documented that a
screening time of two minutes per smear is hardly sufficient for an
experienced cytotechnologist to assess the material as significant or
not significant with a relatively small margin of error.
I am not aware of any other study to determine whether the accuracy of
screening is time-dependent. Clearly, not all cytotechnologists are
equally talented, and while some may be capable of rapidly screening a
large number of smears, others are not. It is not known whether
screening efficiency is reduced by fatigue toward the end of the working
day. This is an area of human performance that has never been
appropriately scrutinized in spite of its potential major impact on
public health. QUALITY
CONTROL
It is generally assumed that every laboratory should or must include a
quality control procedure. No area of pathology is more difficult to
control than cytology There are two aspects of quality control:
screening and interpretation. Screening
As a method of quality control, the Centers for Disease Control require
rescreening of 10% of all negative smears to issue a license for
interstate commerce. The execution of this exercise is left at the
discretion of the individual laboratory. Thus, the rescreened smears may
be consecutive or selected by unspecified criteria. If one considers
that, at most, 10% of all smears show some degree of abnormality, the
results of the rescreening process may be entirely and correctly
negative, and this would still not bear on the performance of the
laboratory. Other approaches have been tried, such as surreptitious
introduction of known abnormal smears into the Routine screening
material. This procedure is time‑consuming; it disrupts the
consecutive numbering of material end introduces false records.
Furthermore, with some experience, astute cytotechnologists recognize
the artifacts; and such smears will be screened with special attention
to detail. While this method may help weed out totally incompetent
screeners, it fails to uncover less obvious deficiencies.
Other methods of quality control include the comparison of cytologic
with histologic findings. This method, while commendable, has its
pitfalls, because the smear is usually the trigger for a biopsy, arid
the biopsy results will be by definition abnormal. On the other hand,
the degree of abnormalities may differ between the smear and the biopsy
specimen, not because of incompetent screening but because of the nature
of the sampling procedure (see below). This approach assumes that the
cytologic and histologic samples will be available for comparison. While
this may be the case in some laboratories, accessibility to such
material is clearly limited for some l~nass‑production cytologic
laboratOries, which may experience difficulties in the retrieval of
histologic material from other institutions.
Another method of quality control of screening is the review of previous
material from women with several consecutive smears. If, with the
passage of turner an abnormal smear is observed, a review of,the
previous smears often uncovers cell abnormalities that were not noticed
Op previous screenings. In my Judgmenlt~this is the best method of
quality control; it requires, however, that the records be computerized
and that all smears be stored for a minimum of five years, and
preferably for ten years. The latter requirement is particularly
burdensome on very large laboratories because of the storage space
required. Furthermore, the regulations pertaining to archives of this
type vary greatly from state to state. Interpretation
The task of rendering the definitive verdict on material screened and
marked by cytotechnologists is a professional prerogative of the
pathologist. While there are some superb interpreters of cytologic
evidence scattered throughout the country, it remains uncertain whether
every pathologist not specifically trained or versed in the difficult
art of cytology is capable of accurate or even prudent interpretation of
cytologic abnormalities. Several years ago the College of American
Pathologists introduced a voluntary self‑testing program of
quality control that included cervical smears. The results were
discussed in a report that failed to attract much attention. The study
documented that clearly abnormal smears may not be recognized as such by
some pathologists and that peer consensus is difficult to reach in
cervical cytology, as in many instances of the histology of precancerous
lesions (see below).
In the state and city of New York, the licensing of laboratories of
cytology is by mandatory examination, an old tradition." The test
currently consists of a set of ten smears from various body sites, with
emphasis on cervical smears. The test is administered to each
cytotechnologist and to the supervising pathologist. Cytotechnologists
are expected to identify abnormalities in the smears and submit the
material to the pathologist for final diagnosis. If the responsible
pathologist fails to achieve average performance, the examination is
repeated; a second failure results in suspension of the director's
license and a requirement for postgraduate experience. With the passage
of time the performance of pathologists has improved significantly.
To the best of my knowledge no competency examination is required in any
other state. The examination once required by the Centers for Disease
Control has been discontinued. California has a legislative limit on the
volume of smears per cytotechnologist, but anecdotal evidence indicates
that this limit is bypassed by moonlighting (Wall
Street Journal, Nov 2, 1987, p 1).
It is quite evident that there is room for improvement in the area of
quality control and licensure of laboratories. Unfortunately, the
implementation of any testing procedure is quite costly and
labor‑intensive. There are many opportunities for
cytotechnologists and pathologists to hone their skills in postgraduate
workshops and seminars. Since evidence of such postgraduate training is
not mandatory for the licensing or operation of laboratories of
cytology, the actual proportion of laboratory personnel participating in
these postgraduate exercises cannot be determined with accuracy. INTERPRETATION
OF CYTOLOGIC EVIDENCE
The interpretation of cervical and vaginal smears belongs to the most
difficult areas of microscopy. While the obvious cases of cancer and
precancerous lesions are usually recognized without major difficulties
by trained observers, smears do not always reflect underlying disease.
This is particularly important in reference to invasive cancer; the
surface of the lesion is often necrotic and covered by debris, and the
smear can fail to reveal obvious cancer cells. Because of blind and
unjustified faith in the accuracy of cervical smears it is often
forgotten that this laboratory test does not replace a careful clinical
examination and that a biopsy must
be done of any visible suspicious cervical lesion. Unfortunately,
many precancerous lesions or cancers may be represented in a smear by
only a few cells with relatively trivial abnormalities. This occurs
mainly in smears obtained without the necessary care but sometimes even
in excellent smears. Many such events were observed in a long‑term
follow‑up study of precancerous lesions of the uterine cervix.25
It must be recognized that not all neoplastic cervical lesions shed
cells in a uniform fashion and that some lesions are difficult to
sample. In the presence of such trivial cytologic evidence, the
abnormalities may be overlooked, with the smear classified as
"negative," or, if the abnormalities are noted, they may be
considered insufficient for diagnosis, with the smear classified as
"atypical. "
The principal goal of cervical smears is not to diagnose overt clinical
cancer but to detect occult
small carcinomas and precancerous abnormalities that may
lead to invasive cancer. Hence, the duty of the pathologist is to
signal the presence of cellular abnormalities, even if they are
relatively minor, and to refer the woman for further scrutiny and
follow‑up. Colposcopy is an important next step in the evaluation
of such patients, as it usually leads to clarification of the nature of
cytologic abnormalities.26 This procedure also requires considerable
training and experience. Furthermore, not all lesions produce
colposcopic abnormalities, particularly if they are located in the
endocervical canal beyond the reach of the colposcope. In such
cases an endocervical curettage should be performed.
It is singularly misleading to obtain a second smear within a few days
or weeks after the first one, either to confirm the previous
results or to clarify the diagnosis in "atypical" cases. For
reasons unknown the second sample may be completely negative in about
609~o of patients with significant neoplastic
lesions.27‑" Thus, in many cases, the first abnormal smear is
considered a 'laboratory and, instead of being referred for further
care, the patient is reassured. This setting is an invitation to
disaster, because some women in the situations described above may ultimately
die of cancer of the cervix, only to be described as victims of a poor
laboratory. PRECANCEROUS
LESIONS OFTHE UTERINE CERVIX AND INVASIVE CANCER
It is often not understood that the ratio of precancerous lesions of the
uterine cervix to invasive cancer is probably of the order of 10 to
l, possibly even higher. In other words, at the most, one in ten
precancerous lesions is likely to progress to invasive cancer if left untreated.
The reasons for the unpredictable behavior of these lesions remain
enigmatic and have been the subject of endless discussions, none of
which have gone beyond speculation or hypothesis. The recent recognition
that human papillomavirus (HPV) infection is closely associated
with precancerous states and invasive carcinoma in the uterine cervix
has raised the hope that the identification of the virus type may
have prognostic valued' Thus, precancerous cervical lesions
associated with HPV types 16, 18, 31, 33, and other
"high-risk" types are thought to be more likely to progress to
invasive cancer than lesions associated with HPV types 6 and ll,
commonly observed in condylomata acuminate. This evidence must be considered
tentative. It has been recently documented that in the normal, sexually
active population there is a very large pool of infected individuals,
currently estimated at a minimum of 10%, and probably much higher.32
About half of the pool includes carriers of the "highrisk"
types of virus. It has also been shown that the morphologic identification
of early neoplastic changes does not depend on the viral typed although
this observation is the subject of a debate." The association of
invasive cancer with HPV type It has also been reported.35 The presence
of HPV types 6 and It has been documented in lesions, such as the giant
condyloma of Buschke‑Loewen stein, that behave in a locally
aggressive fashion and in condylomas of the urinary bladder, which are
sometimes extremely difficult to eradicate.36
For many years attempts have been made to separate precancerous lesions
that are more likely to progress to invasive cancer from those that
are less likely to do so. This resulted in the classification
of precancerous lesions into two groups: (1) dysplasia, less likely or
unlikely to progress, and (2) carcinoma in situ, more likely to
progress. The criteria for this separation have never been properly
spelled out, and the issue was discussed at length in previous publications.~37
It has been shown in previous studies that under conditions of uniform,
careful follow‑up there are no significant behavioral
differences between these two groups of lesions.25 Furthermore,
there is excellent evidence that even competent pathologists cannot
agree with each other on the classification of these lesions based
on ample histologic evidence, let alone smears.
The concept of cervical intraepithelial neoplasia (CIN) introduced by
Richard included all precancerous events, regardless of histologic
type, in one group of precancerous lesions requiring colposcopy and
biopsy evaluation and treatment, regardless of morphologic findings. THE
PAP SMEAR: THE TRIUMPHS
There is ample evidence that the cervical smear has contributed in
a remarkable fashion to the prevention of invasive carcinomas
of the uterine cervix. Numerous studies clearly document a statistically
valid drop in the incidence and mortality rates of this diseased"
Particularly impressive are the data from the early screening programs
in Louisville, ds British Columbia,46'7 and, more recently, the Nordic
European countries.^5' The latter studies are valuable because they
underwent a thorough evaluation by competent epidemiologists. With
the passage of time, usually after five or more years, a statistically
significant reduction in mortality from cervical cancer in the screened
population has been observed. The reduction varies in scope
depending on the frequency of smears, the intensity of rescreening, the
competence of the team, and the effectiveness of the follow-up
procedure.59-55
It can be stated without any equivocation that the cervical smear
is an effective tool of cancer detection, perhaps
the only effective cancer screening test known today. What has not
been stressed adequately is the fact that screening
for carcinoma of the uterine
cervix has not eradicated the disease in any of the populations studied
to date. There are multiple reasons for the failure of the
screening systems, and the will be analyzed. THE
PAP SMEAR: THE TRAGEDY
It is quite evident from the abo summary that the complex cervical cab.
cervix detection system is prone to failure at all levels, starting with
the inn' clinical examination, continuing with the taking of the smear
sample and la oratory errors in screening and into pretation, and ending
with the clinician’s failure to understand the reps or take
appropriate action and, in Sol instances, with the patient's failure
follow the guidance of the physician. Accuracy
of the System
In the absence of controlled clinical trials, the failure rate of the
cervix cancer prevention system is extremely difficult to evaluate. In a
thoughtful study, Martins evaluated the reasons for s; tem failure in
two communities, one which has a very high rate of cytologic screening,
estimated at 95%.or high and the other without a screening program. Data
were compiled for the y 1967 as a case study procedure for patients in
whom stages 1 through invasive cervical carcinoma develop microinvasive
carcinomas were excluded. There were 67 carcinomas in ' Diego County,
where the screening, encouraged, resulting in a prevalence rate of 5/100
000, and there were n carcinomas in adjacent Imperial County, resulting
in a prevalence rate 11/100 000. With medicolegal immu assured, the
failures could be attribute in an overlapping manner to patient error,
to physician error, and laboratory error. Similar observations were
ported from Sweden by Rylander'6 from Italy by Ceccini et al.57
Patient Error.𢰦patient
error tabulated by Martins consisted ma of failure to undergo
follow‑up am medical examinations (34 of 58 p tients), delay in
visits to physicians' flees while symptomatic (17 of 58 patients), and
refusal of diagnostic measures (seven of 58 patients). Wied et ,
examining the huge repository of c. puterized records at the University
Chicago, also observed a very high r of failure of patients to return
for ann screening. Obviously, either the won were not aware of the need
for ann return or they were prevented from turning for other reasons.
Physician Error.𦳀artin
attrib ed 42 of 76 invasive cancers to physic error, most commonly
failure to p form a pelvic examination and obtain cervical smear (30
patients). Other rors consisted of failure to act on a port of abnormal
cytologic findings a other events, such as untrained phy clans reading
smears in their Offia Rylander'6 also pointed out that the agnosisof
"dysplasia" was often considered inconsequential by uninformed
physicians, who failed to take the action most consistent with
patients" welfare.
Laboratory Error.鎸n
Martin's study, smears were available for review in only 13 of 76 cases.
Ten results were incorrectly reported as negative. The literature is
replete with reports of failures of the cytology laboratory to dings
nose invasive cancer.s‑5 One study' re ported a cumulative
false‑negative error rate for invasive cancer of about 50%. In the
same study, the rate of screening errors for precancerous lesions was at
least 28%. Another source of information about the false‑negative
screening rate is a series of reports documenting that two cervical
smears obtained simultaneously reduced the false‑negative
laboratory error rate by at least 20% for precancerous lesions.65 67
Regardless of the percentages, it is clear that the error rate of
cytologic screening for precancerous lesions and invasive cancer of the
uterine cervix is quite substantial and may remain so regardless of the
remedial actions suggested below. REMEDIES Dissemination
of Information
Some limitations of the screening system, of the current understanding
of cancerous events in the uterine cervix, and of the constraints and
obligations imposed on pathologists and clinicians have not been widely
recognized. There are still some or perhaps many members of the medical
profession directly or indirectly involved in screening for cervical
cancer who believe that the Pap smear is infallible, that "dysplasia"
is a benign disease that will go away, and that carcinoma in situ is a
life‑threatening disorder that nothing will cure short of a
hysterectomy. None of these are true.
It is quite evident that the cervix cancer control system accomplished a
great deal and saved the lives of a great many women, but it is also
capable of abysmal failures. If the detection system is to work better,
the first step should be to inform women and their physicians about the
current status of the system and make them aware of its many weak
points. The improvements must take place in every link of the chain of
events, beginning with increasing women's awareness of the need for
repeated examinations and encouraging them to insist on clinical and
laboratory excellence. Clinical
Component
Obtaining an adequate and representative cervical smear should be only
one part of the clinical examination. An accurate history of sexual
exposure may be as valuable as any other component of the system.
High‑risk factors include a multiplicity of sexual partners and
onset of sexual activity at an early age. While it is not customary
today to ask these questions as part of the general examination, the
identification of women at risk, regardless of age, is helpful not only
in the detection of cervix cancer but also in identifying women at risk
for sexually transmitted diseases, including the acquired
immunodeficiency syndrome. Other factors of importance are the obstetric
history, history of previous lesions, treatment received, etc. This
information must accompany the smears to come to the attention of
laboratory personnel. Because the Pap test is unlike any other
laboratory procedure and is dependent only on human judgment, the
interpretation of the evidence may depend to a significant extent on the
clinical data.
The clinical status of the external genitalia, the vagina, and the
uterine cervix plays a significant role in the cancer detection system.
For example, the presence of condylomata acuminate is suggestive of
infection with HPV and indicates that the patient should be classified
in the high‑risk category for cervical neoplastic disorders. The
presence of other sexually transmitted diseases, such as gonorrhea,
syphilis, or Chlarnydia
trachornatis infection, also suggests that high‑risk factors
are at play. Inspection of the vagina and the cervix may reveal
abnormalities that justify clarification by biopsy regardless of the
results of the smears. The rate of failure of cytologic examination in
invasive cancer of the cervix is high, and diagnosis by biopsy is
essential anyway. In my judgment, it is better to take one biopsy
specimen too many than to miss one invasive cancer.
Postmenopausal Patient.𦝁oo
often the failure of the system is based on the erroneous notion that
women past menopause do not require cervical examination or smears. In
fact, in this age group the cervix cancer detection system fails with
surprising frequency. Women past the age of 50 in whom invasive
carcinoma develops often do not receive appropriate cancer detection
care. This group of women may also benefit from a vaginal pool smear,
which may occasionally help in the detection of occult carcinomas of the
endometrium, tube, or ovary.20 Recently, Mandelblatt and Fahs~ suggested
that screening of low‑income elderly women may be particulary
cost‑effective.
Frequency of Screening.娋Because
of smear failure and to ensure optimal performance of the system, at
least three~sequential annual smears should be obtained to rule out
cervical disease.
This is the recommendation of the American Medical Association, the
American Cancer Society, and the appropriate bodies of gynecologists
(New York limes, Jan 7, 1988,
p B13),. but one wonders to what extent the medical profession at large
and the public are aware of this prudent recommendation. From the
evidence presented above, it is clear that many if not most women do not
follow this recommendation, particularly if they do not receive annual
obstetrical or gynecologic care or cannot afford it. Obtaining two
smears at every examination will clearly enhance the accuracy of the
systems but will also increase the cost substantially. Cytology
Laboratory Component
Adequacy of Smears.𢰦erhaps
the most important first step in the function of a cytology laboratory
is to recognize inadequate cervical smears and report them to the
providers. Recently, VooiJs et ale stressed the importance of the
presence of endocervical cells as evidence of smear adequacy. This
remains somewhat controversial, because in some women endocervical cells
cannot be secured, and in others the presence of the cervical mucus in
smears is just as reliable a sign of smear excellence.28 Still, in a
woman of childbearing age, some evidence should be present to indicate
that sampling of the uterine cervix was adequate. While it is not a
pleasant task and one that is bound to create difficulties between the
provider, the laboratory, and the third‑party payer, reporting
inadequate smears is nonetheless an important first step in ensuring
laboratory performance. Depending on the skill of the provider, from 10%
to 20% of routine cervical smears are inadequate, in my experience.
Quality Control.妔Rescreening
of 10% of smears is a very poor qualitycontrol measure. Rescreening of
all smears or at least those that come from high‑risk patients is
an advisable but costly step. Computerized record keeping for the
comparison of smear results with the results of follow‑up, such as
colposcopy and biopsy findings, is equally important. A recall system
that requests follow‑up information on each abnormal smear would
go far in reminding providers about the need to take the necessary steps
to clarify the cytologic diagnosis.
Whether an examination of cytoscreeners and pathologists should be
mandated in states other than New York is a political and fiscal issue
that cannot be addressed here. Results from New York states' strongly
suggest that this measure would improve laboratory performance. Another
quality‑control measure that should be considered is limitation of
the volume of smears to be screened by a single individual to manageable
proportions geared to the individual's talents and capabilities. Qualifications
of Laboratory Personnel
Cytotechnologists, like registered nurses, are highly trained
individuals who perform an extremely difficult task. The fiscal rewards
for these people are generally modest, and this accounts to a
significant extent for overtime work and for "moonlighting"
(Wall Street Journal, Nov 2, 1987, p 1). Many outstanding
cytotechnologists "burn out" and leave the profession for
easier and more rewarding fields. Furthermore, schools for
cytotechnologists, which at one time benefited from federal support, are
now self‑sustaining and often charge significant tuition. As a
consequence, the pool of highly qualified candidates is drying up. If
high‑volume I jaboratories were to reduce the work load to ho more
than 50 smears per day per cytotechnologist, a measure that might reduce
the error rate, an acute shortage of cytotechnologists would occur
immediately.
Pathologists highly qualified in the interpretation of cytologic
evidence are also in short supply. The American Board of Pathology
recently recognized the need for subspecialty training in topathology
and will conduct its first qualifying examination in 1989. The number of
fellowships in this area is still small but should grow with time. This
type of training is likely to reduce some of the problems with the
interpretation of cytologic evidence that are known to exist in many
laboratories.~°" Reporting
Results of Smears
Papanicolaou was not a pathologist, and he was always concerned about an
inappropriate cytologic diagnosis of cancer. To avoid difficulties with
tissue pathologists, he proposed a series of classes ' to describe
smears, with class I being normal and classes IV and V indicating
cancer. A class II smear indicated an "atypia" of a benign
nature, and class III indicated a suspicion of cancer. This system of
reporting is still m use in many laboratories but it is clearly
deficient for abnormal smears unless it is accompanied by a verbal
description of the findings. Because there is a great deal of confusion
about the clinical significance of the terms dysplasia and carcinoma in
situ, the terminology suggested by Richart, ~ CIN, is likely to bring a
more active approach to the evaluation and treatment of precancerous
lesions of the uterine cervix. By agreement between the laboratory and
the provider of smears, further steps to be taken in the investigation
of patients mayalso be suggested. While some physicians encourage such
reporting, others are against it as an infringement of the laboratory on
their clinical judgment. Unfortunately, sometimes the judgment is not
based on a sound understanding of the current knowledge of the disease
processes, with resulting undertreatment or overtreatment of patients. After
care of Patients With Abnormal Smears
There is no unanimity among gynecologists and pathologists about when
further evaluation of the uterine cervix is mandatory. Some suggest only
cytologic follow‑up for women with a cytologic diagnosis of
"mild dysplasia" or grade I CIN. For a number of reasons,
including timid interpretation of smears, nonrepresentative smears, and
the failures of follow‑up smears discussed above, it is my view
that such women should be referred for colposcopic evaluation. Clearly,
all women with higher grades of abnormality must also undergo colposcopy.
In fact, the term atypia (class II smears) often obscures significant
abnormalities of the uterine cervixes I have witnessed the development
of invasive cervical cancer following reports of "atypical"
smears, and many such women could benefit from colposcopy. Colposcopy
Colposcopy is not within my area of expertise, but in experienced hands
this is an excellent method of evaluation of the uterine cervix.
Colposcopically guided biopsy specimens often clarify scanty or obscure
cytologic findings and give an assessment of the size, location and
scope of lesions. In fact, an optimal cancer detection system should
probably consist of a cervical smear and a colposcopy, since colposcopy
alone is an insufficient cancer detection method. In the presence of a
suspicious or positive smear and the absence of colposcopic findings, an
investigation of the endocervix or even the endometrium may be
necessary. What
Can Be Expected?
The cervical cancer detection system is meant to protect women from a
potentially deadly disease: invasive carcinoma of the uterine cervix.
The goal of total or near‑total eradication of this disease is
theoretically achievable, as Marshall'2 showed many years ago in a
stable medical community. One to three competently obtained and
interpreted negative annual cervical smears offer a significant level of
protection for a period of several years but are not a guarantee of
nonoccurrence of cervical cancer. For women between the ages of 35 and
64 years, the level of protection is estimated to be 93.5% for annual
screening, 83.6% for screening every fifth year, and 64% for screening
every ten years.
The cost of screening was recentl calculated by Eddy for screening fre
quencies from annually to once ever: ten years. The calculation did not
in elude the cost of office visits but did include the cost of treatment
of various types of lesions and took into consider ation certain
assumptions about the efffi cacy of the smears. The cost of annual
screening of 100 000 women for 50 years (from 20 to 70 years of age) was
calculated to be over $29 million, or about $5.80 per woman per year.
From these data the cost of 50 years of annual screening of the 50
million adult American women would be $14.5 billion, or $290 million per
year. This figure may be uncharac-T teristically low because the
costs of Pap smears will increase once the quality control measures
mandated by the newly passed amendment to the Clinical Laboratory
Improvement Act (HR 5471) are implemented. Screening every third or
fifth year would cost significantly less but would offer a lower degree
of protection. Hence, screening for cervical cancer becomes a public
health policy that must compromise between the expected and the
reasonable. The first step, which will ultimately lead to the
implementation of this policy, is to make known to the public the
benefits and limitations of the system so that the Pap smear will not
again become the prey of sensational press reports that spread fear and
distrust of a fundamentally sound cancer detection system. How
to Do It?
The cervical cancer detection system, as it has developed over the years
in the United States, lacks cohesiveness and cannot be easily regulated
by congressional or administrative fiat. A new beginning is clearly
needed. Although we do not wish to separate women from their physicians,
it nevertheless appears reasonable to propose the creation of
well-wo1nen centers, which would offer a broad variety of health care
services to asymptomatic women of all ages at moderate cost. Some of the
problems with securing cervical and vaginal smears could be better
controlled if paramedical persons were appropriately trained. It is not
inconceivable that other services, such as mammography or even prenatal
care, could be combined in such health centers; this could benefit women
who are unable to afford private care. The option of creating
outstanding central laboratories to service such centers under carefully
supervised and controlled conditions should also be considered.
Many of the recommendations in this article have recently been
incorporated.
|
|