BASAL CELL CARCINOMA:
IMPORTANCE OF HISTOLOGIC DISCONTINUITIES
IN THE EVALUATION OF RESECTION MARGINS.
Seidman JD, Berman JJ, Moore GW.
http://www.netautopsy.org/basalcel.htm
Send comments and correspondence to:
George.Moore4@med.va.gov
See also:
http://www.medparse.com/gwmcv .............
U. S. Government Work, uncopyrighted, published as:
Seidman JD, Berman JJ, Moore GW.
Basal cell carcinoma: importance of histologic discontinuities
in the evaluation of resection margins.
Mod Pathol. 1991 May;4(3):325-330.
Abstracted in Yearbook of Pathology
and Clinical Pathology, eds., Gardner WA jr, Bennett BD, Cousar JB,
Garvin AJ, Worsham GF, St. Louis: Mosby, pp. 189-190, 1993.
PMID: 2068058.
PubMed Entry
Full Text of Article:
http://www.netautopsy.org/basalcel.htm
ABSTRACT.
Pathologists frequently need to judge whether basal cell carcinomas
have been excised adequately. Traditionally, excision adequacy is assessed
by looking for the presence of tumor at the margins of resection.
This time-honored activity has questionable value, since it has been
demonstrated that the majority of tumors with positive margins
do not recur, and a substantial minority of tumors with negative margins
do recur. It is proposed that excision adequacy can be evaluated by
considering the pattern of tumor growth. Tumors composed of widely dispersed
nests need wider margins than tumors that grow as tight clusters
of tumor nests, and this assertion can be evaluated statistically.
A morphometric study of 28 basal cell carcinomas (BCCs) was performed,
in which the distribution of tumor cell nests seen in cross-section
was analyzed. The average distance from the center varied greatly
(272 microns up to 2273 microns) among these tumors. Standard deviations
were calculated from distances between the tumor center to each nest
within the tumor, and one-tailed Student t-tests were used to obtain
90%, 95% and 99% confidence limits for distances beyond which
no additional tumor nests are expected. These distances, in tumor radii,
ranged from 0.8 to 1.9 and from 1.03 to 4.89, for 90% and 99%
confidence limits, respectively.
Conventional methods used to determine margin adequacy do not account
for the discontinuous appearance of BCC in histologic sections.
This theoretical model demonstrates that an alternate way of assessing
excision adequacy can be achieved with a statistical analysis
of the pattern of tumor growth, rather than looking for absence of tumor
at the resection margin. Further studies are needed to determine whether
a statistical evaluation of excision adequacy has clinical benefit.
Key Words: Basal cell carcinoma, resection margins, tumor growth
INTRODUCTION.
Basal cell carcinoma (BCC) is the most common malignant tumor in man,
but it has no single standard of treatment, and recurrence rates can be
distressingly high even in lesions with `negative' surgical margins.
Furthermore, there is no common understanding among pathologists and surgeons
as to the meaning of the commonly used terms, `free margin,' `close margin',
or `complete evaluation' of margins (1).
In practice, these terms are used inconsistently.
The underlying assumption in all tumor margin examination is that
tumors grow contiguously. If there is no tumor at any of the sampled
surgical margins, then the tumor is said to be `completely excised,'
and therefore there should be no local recurrence of tumor. A cross-section
of basal cell carcinoma typically consists of discrete nests
of varying sizes. Figure 1 is a graphic
conceptualization of a cross-section of basal cell carcinoma.
A central tumor mass sends out processes and on cross-section produces
a collection of nests separated by stroma. The arrow
(Figure 1) points to a focus where no tumor
is present in the cross-section. However, tumor is present on either side
beyond the plane of sectioning. This `skip' phenomenon in tumor growth
may be responsible for the high recurrence rate of tumors with apparently
negative margins.
The examination of margins in general, and of basal cell carcinomas
in particular, is subjective and varies greatly between observers
(1). Furthermore, the distribution
of tumor cell nests can vary from tumor to tumor, and this variation
should bear on the pathologist's evaluation of margins.
Figure 2a and
Figure 3a show two extremes of tumor nest
patterns in basal cell carcinoma.
In Figure 2a,
the tumor nests grow tightly together. A pathologist might consider
that a 4 mm margin around the cluster of nests represents
an adequate excision. In Figure 3a,
the nests are widely separated. A pathologist might consider
that a 4 mm margin beyond the furthest nest, although technically free
of tumor, is an inadequate excision. How does the pathologist convey
to the surgeon that a wider excision appears indicated even though
the margins are technically adequate? If the surgeon widens the excision,
how does the pathologist know when a "safe" margin is reached?
The pathologist's ability to guide the treatment of the patient
is currently limited because there are at present no objective features
that describe and predict tumor growth based on measured samplings of tumor.
In the present report we examine and quantify the distribution of tumor nests
in 28 basal cell carcinomas by sampling the distribution of nests in a
representative cross-section of the original excision. Furthermore,
using an index of nest cohesiveness derived from this analysis,
an objective prediction is made of the likelihood that nests will appear
beyond an arbitrary point (e.g., the surgical margin). This study
demonstrates that judging adequacy of excision by using measurable
references in the pattern of tumor growth, independent of any observations
made at the excision margin, is theoretically possible.
MATERIALS AND METHODS.
Eighty-nine consecutive basal cell carcinomas listed in the files
of the Baltimore Department of Veterans Affairs Medical Center over
a 6 month period in 1989 were examined. Criteria for inclusion
in the study were: (1) the tumor appeared to be completely or almost
completely excised; (2) at least one well-oriented section was available,
showing the skin surface and deep portions of the tumor; and (3) most
or all of the tumor had grown in nests, islands, or cords, so that for the
majority of such structures, the center of the structure could be estimated
with reasonable accuracy.
A well-oriented section of each tumor that met these criteria
was photographed in its entirety at a magnification of 25x, and prints
were assembled to reconstruct the complete cross-section. An estimated
central point of each tumor was chosen as the origin, and x and y axes
were drawn so that the skin surface intersected the positive half of the
y axis, and the x axis was parallel to the skin surface. A center point
was chosen for each nest and (x,y)-coordinates (horizontal and vertical
distances from the origin) were recorded. When the center of a nest could
not be estimated due to irregularity of the nest, several points in the nest,
each representing the center of a subnest, were plotted and the center
was obtained as the arithmetic mean of these points. Data were entered
on the File Manager (FileMan) database management system of the Department
of Veterans Affairs, a public-domain software package
(2).
All statistical calculations were made with public-domain programs
written in American National Standard MUMPS. For each tumor a `true center'
of the point distribution was taken as the arithmetic mean of the x and y
coordinates. The Euclidean distance of the center of each nest from the
`true center' of the tumor was calculated. Mean, standard deviation,
and one-tailed Student confidence limits (90, 95, 99%) were calculated
for the distances from the `true center' to the nest centers of each tumor.
The diameter of each tumor was estimated as the maximum inter-nest distance,
and statistics were normalized to half the diameter (i.e., radius)
of each tumor.
RESULTS.
Twenty-eight basal cell carcinoma specimens met
the inclusion criteria for the study. Data are summarized in
Table 1,
and complete analyses for the two tumors shown in
Figure 2 and in
Figure 3.
are given in
Table 2.
The number of nests examined per tumor ranged from 3 to 141. The radii of the tumors ranged from 540 microns to 4285 microns. The average distance from each nest to the `true center' ranged from 272 microns to 2273 microns. Ninety percent confidence limits represent the distances from the `true center' of the lesion to a radial point beyond which there is a 10% chance of finding additional tumor. Ninety percent confidence limits ranged from 533 to 3531 microns (0.80 to 1.90 radii); 95% confidence limits ranged from 594 to 3895 microns (0.89 to 2.51 radii); and 99% confidence limits ranged from 690 to 4592 microns (1.03 to 4.89 radii).
Analysis of a BCC with high nest cohesiveness is shown in
Figure 2 and
Table 2.
The average distance from 42 nest centers to the `true center' is 377
microns, and the standard deviation of this set of distances is 129 microns.
Based on these data and the Student t distribution, the 90%, 95%, and 99%
one-tailed confidence limits are 545, 594, and 690 microns from the center,
respectively. The radius of this tumor is 540 microns, so that the same
confidence limits are equivalent to 1.01, 1.10 and 1.28 radii,
respectively (concentric circles, shown in
Figure 2C).
If one wishes to achieve clear surgical margins with 90% certainty,
the excision should be 545 microns from the center. With this size excision,
there is a 10% probability that residual tumor was left in the patient.
To achieve clear surgical margins with 95% or 99% certainty
(5% or 1% probability that some tumor will be left in the patient),
the excision should be widened to 594 or 690 microns from the center,
respectively.
The analysis of a BCC with low nest cohesiveness is shown in
Figure 3 and
Table 2.
The average distance from 4 nest centers to the `true center' is 651 microns,
and the standard deviation is 449 microns. As above, the 90, 95, and 99%
one-tailed confidence limits are 1386, 1707, and 2690 microns from the
center, respectively . The radius of the tumor is 1034 microns, so that
these confidence limits are equivalent to 1.34, 1.65, and 2.60 radii,
respectively (concentric circles in Fig. 3b). To achieve clear surgical
margins with a 90% certainty (10% probability that residual tumor will be
left in the patient), the excision should be 1386 microns from the center.
To achieve clear surgical margins with 95% or 99% certainty (5% or 1%
probability that some tumor will be left in the patient), the excision
should be widened to 1707 or 2690 microns from the center, respectively.
These findings predict smaller distances for adequate excision
in tightly-nested BCC's compared to BCC's composed of scattered nests.
DISCUSSION.
The reported recurrence rates of BCC treated with standard techniques
vary widely. Crissey's review showed 92.6% to 95.5% cure rates for surgery,
radiation or cautery, and electrodessication (3).
Some authors believe that recurrence rarely exceeds 5%
(4); however, in sites where minimal excisions
are preferred (particularly certain areas of the face), recurrences
of 10% to 25% are common. Once a BCC recurs, it becomes more difficult
to eliminate than a newly presenting lesion.
Menn (5) achieved only 50% cure
of recurrent BCCs treated and retreated with curettage and
electrodessication, surgery, or radiation.
Surprisingly, only a minority of patients with histologically
documented positive margins will experience a recurrence.
DeSilva and Dellon (6) followed 38 patients
with positive margins and only 37% recurred. Pascal
(7) found 33% recurrence with positive margins,
12% recurrence when BCC was within one high power field of the margin,
and 1.2% recurrence when the margins were free (excluding multifocal
lesions). Gooding (8) reported
a 35% recurrence rate and Shanoff (9)
reported 67%, both with positive margins. Sarma
(10) studied 43 immediate reexcisions
of BCCs with positive margins. All tissue was processed for examination,
and only three had residual tumor. It was suggested that tumor cells
at the operative site may be devitalized.
Statistics show that a significant fraction of tumors
with negative margins will recur. Lang (11)
studied ten recurrent tumors, six of which had negative margins
with the initial excision. Hauben (12)
found a recurrence rate of 25.6% with positive margins and 22.8%
with negative margins.
Key to any discussion of BCC is an appreciation that the majority
of tumors with proved positive margins do not recur, and that a sizable
minority of tumors with negative margins do recur. Thus, the biological
significance of the histologic evaluation of resection margins of BCC
is questionable and must be critically evaluated.
Proponents of the Mohs chemosurgical and fresh tissue techniques
claim to overcome the weaknesses inherent in margin sampling by examining
the entire resection margin. This is done by tangential sectioning of all
deep and lateral margins, layer by layer, until negative margins
are achieved. Five-year cure rates of 96% to 99.6% have been reported
for initial excisions (13,
14, 15,
16, 17,
18, 19),
as well as 96% to 97% cure rates for recurrences
(14, 16,
18). Others have reported cures
for recurrent BCCs of only 75% to 84% by Mohs technique.
The Mohs technique is practical only for small BCCs due to
difficulties of mapping and sectioning the entire margin of large lesions.
Implicit in all current methods of margin evaluation
is the assumption that there are no discontinuities in tumor growth
(i.e., the tumor grows as a solid object and cannot disappear
and reappear on the other side of a margin section). This supposition,
in our opinion, is unsubstantiated. Madsen (20)
claimed to show that all nests in a case of BCC are contiguous
in three dimensions, but this has not been corroborated by others.
Dzubow (21, 22,
reported two cases in which negative Mohs layers were followed by
deeper positive layers. Imayama (23,
24), in an ultrastructural study
of superficial BCCs, suggested that nests seen in cross section represent
the proximal aspects of the tumor, and that the tumor itself extends
through the epidermis as a single cell layer which is not recognizable
histologically. Imayama's view supports a unicentric origin of BCC,
where connections between nests may be only a single cell layer,
and thus not recognizable as tumor cells on routine sections.
Although tumor cell nests grow contiguously, strands may, over time,
degenerate in some areas while nests proliferate in other areas,
producing discontinuities. Furthermore, iatrogenic skip areas are
frequently produced due to previous biopsy (16,
21, 22,
25).
We believe that margin adequacy can be assessed objectively
by an analysis of the growth pattern of the tumor. Our data establish
confidence limits for the location of tumor cell nests in a series of BCCs.
These limits are based on the distribution of distances of all nests
from the mathematical center of the tumor and the assumption that
nest distances are sampled from a normal distribution. The tumor with high nest dispersal (low cohesiveness) requires a wider excision than the tumor with lower dispersal to achieve a similar probability of complete excision. This finding supports our contention that BCCs that have widely separated nests require wider excisions than BCCs of similar apparent size with more closely-spaced nests. The importance of this concept is underscored by our prediction that the tumors most likely to give false negative margins (tumors with high dispersal, because they contain wide spaces without tumor) require the widest excisions. Although the absolute radius of the required excision varies widely among the 28 tumors studied (594 to 3895 microns, for 95% confidence), when normalized to the tumor radius, a fairly narrow distribution of numbers is found (0.89 to 2.51 radii).
We have made several assumptions in our analysis of the 28 tumors in this study: (1) The cross-section studied was representative of the entire tumor (i.e., any other tumor cross-section would have shown a similar distribution of nests). (2) The section studied was through the center of the tumor (i.e., our calculation of the `true center' of the tumor was valid). (3) The distribution of distances of the nest centers to the `true center' approximates a normal distribution. An additional limitation of this analysis is that the nesting patterns often vary within individual tumors, and therefore analysis of one representative cross-section may not adequately reflect the whole tumor. The proof that our analyses are valid awaits empiric validation. Specifically, will we see clinical recurrences of BCC in instances when our analyses judged the margins to be `inadequate'? In the cases where our analysis indicates that margins were safe to a 95% or 99% confidence limit, will we see no recurrences?
The assessment of surgical resection margins is a daily ritual for most surgical pathologists. Excision adequacy is traditionally assessed by looking at the resection surface and determining whether there is tumor at the margin. We question the validity of this approach because it is not a reliable predictor of recurrence and because it fails to take into consideration the growth pattern of the tumor proximal to the resection margin. In this report, we offer a theoretical alternative to margin sampling for determining tumor excision adequacy. This may be a cumbersome method to use with every BCC. However, a pathologist can make a rapid visual assessment of nest distribution and decide whether a tumor is very cohesive, highly dispersed, or somewhere in between. Such assessments may eventually be useful additions to pathology reports if this thoeretical model is found to have predictive power in the clinical setting. Clinical studies are needed to determine the predictive power of a statistical assessment of BCC margins.
REFERENCES.
1. Abide JM, Nahai F, Bennett RG.
The meaning of surgical margins.
Plastic Reconstr Surg. 1984;73:492.
2. Davis RG.
FileMan: A User Manual.
Bethesda, MD: National Association of VA Physicians. 1987;:.
3. Crissey JT.
Curettage and electrodessication as a method of treatment
for epitheliomas of the skin.
J Surg Oncol. 1971;3:287.
4. Koplin L, Zarem HA.
Recurrent basal cell carcinoma: a review concerning the incidence,
behavior and management of recurrent basal cell carcinoma,
with emphasis on the incompletely excised lesion.
Plastic Reconstr Surg. 1980;65:656.
5. Menn H, Robins P, Kopf A, Bart R.
The recurrent basal cell epitheliomas.
Arch Dermatol. 1971;103:628.
6. DeSilva SP, Dellon L.
Recurrence rate of positive margin basal cell carcinoma:
results of a five-year prospective study.
J Surg Oncol. 1985;28:72.
7. Pascal RR, Hobby LW, Lattes R, Crikelair GF.
Prognosis of "incompletely excised" versus "completely excised"
basal cell carcinoma.
Plast Reconstr Surg. 1968;41:328.
8. Gooding CA, White G, Yatsuhashi M.
Significance of marginal extension in excised basal-cell carcinoma.
N Engl J Med. 1965;273:923.
9. Shanoff LB, Spira M, Hardy SB.
Basal cell carcinoma: a statistical approach to rational management.
Plast Reconstr Surg. 1967;39:619.
10. Sarma DP, Griffing CC, Weilbaecher TG.
Observation on the inadequately excised basal cell carcinomas.
J Surg Oncol. 1984;25:79.
11. Lang PG, Maize JC.
Histologic evolution of recurrent basal cell carcinoma
and treatment implications.
J Am Acad Dermatol. 1986;186.
12. Hauben DJ, Zirkin H, Mahler D, Sacks M.
The biologic behavior of basal cell carcinoma: analysis of recurrence
in excised basal cell carcinoma. Part II.
Plast Reconstr Surg. 1965;69:110.
13. Breuninger H.
Histologic control of excised tissue edges
in the operative treatment of basal-cell carcinomas.
J Dermatol Surg Oncol. 1984;10:724.
14. Cottel WI, Proper S.
Mohs' surgery, fresh-tissue technique: our technique with a review.
J Dermatol Surg Oncol. 1982;8:576.
15. Mohs FE.
Chemosurgical treatment of cancer of the eyelid:
a microscopically controlled method of excision.
Arch Ophthalmol. 1948;39:43.
16. Mohs FE.
Chemosurgical techniques.
Otolaryngol Clin North Am. 1982;15:209.
17. Rigel DS, Robins P, Friedman RJ.
Predicting recurrence of basal cell carcinomas treated by
microscopically controled excision: a recurrence index score.
J Dermatol Surg Oncol. 1981;7:807.
18. Roenigk RK.
Mohs' micrographic surgery.
Mayo Clin Proc. 1988;63:175.
19. Wolf DJ, Zitelli JA,
Surgical margins for basal cell carcinoma.
Arch Dermatol. 1987;123:340.
20. Madsen A.
Studies on basal-cell epithelioma of the skin.
Acta Pathol Microbiol Scand. 1965;177(suppl):9.
21. Dzubow LM.
Chemosurgical report: recurrence (persistence) of tumor
following excision by Mohs surgery.
J Dermatol Surg Oncol. 1987;13:27.
22. Dzubow LM.
False-negative tumor-free margins following Mohs surgery.
J Dermatol Surg Oncol. 1988;14:600.
23. Imayama S, Yashima Y, Higuchi R, Urabe H.
A new concept of basal cell epitheliomas based on the
three-dimensional growth pattern of the superficial multicentric type.
Am J Pathol. 1987;128:497.
24. Imayama S.
Morphologic basis for adequate surgical excision
of basal cell epithelioma.
Arch Dermatol 1988;124:1715.
25. Wagner RF Jr, Cottel WI.
Multifocal recurrent basal cell carcinoma following
primary tumor treatment by electrodessication and curettage.
J Am Acad Dermatol. 1987;17:1047.
FIGURE LEGENDS.
Figure 1. Graphic representation of the growth
of a basal cell carcinoma. A plane through the tumor represents
the view of basal cell carcinoma seen in a typical tissue section
and is composed of scattered nests.
Figure 2A. A nodular basal cell carcinoma
with tightly packed nests.
Figure 2B. Schematic representation of measurements
taken for a BCC. Each cross represents the center of a nest.
The geometric center of these points (`tumor center') is calculated
by taking the mean of the x and y coordinates of every nest
(a few coordinates of which are shown). The oblique lines
represent distances of the nest centers from the tumor center.
These distances form the data set from which
standard deviations and confidence limits are calculated.
Figure 2C. Graphic representation of the same tumor
with concentric circles, indicating confidence limits of tumor margins
based on a statistical analysis of the distribution of nests.
To attain a margin with 99% confidence that no nests lay beyond the margin,
the tumor should be excised 1.28 radii from the tumor center.
Figure 3A. A basal cell carcinoma
with widely dispersed superficial nests.
Figure 3B.
Graphic representation of the same tumor with concentric circles
indicating the confidence limits of tumor margins based on
a statistical analysis of the distribution of the nests.
To attain a margin with 99% confidence that no nests lay
beyond the margin, the tumor should be excised 2.6 radii
from the tumor center.
TABLE TITLES.
TABLE 1. NEST DISTRIBUTIONS IN 28 BASAL CELL CARCINOMAS,
WITH ONE-TAILED STUDENT t CONFIDENCE LIMITS.
TABLE 2. NEST DISTRIBUTIONS FOR TWO BASAL CELL CARCINOMAS
(CORRESPONDING TO THE TUMORS SHOWN IN FIGURES 2 AND 3)
WITH ALL NEST-TO-TUMOR CENTER DISTANCES DISPLAYED.
Last updated: 9/11/2005, by G. William Moore, MD, PhD.