Seidman JD,Berman JJ, Yetter RA, Moore GW.
Multiparameter DNA flow cytometry of keratoacanthoma.
Anal Quant Cytol Histol 14(2):113-119, 1992
MULTIPARAMETER DNA FLOW CYTOMETRY ANALYSIS OF KERATOACANTHOMA
Jeffrey D. Seidman, M.D. (1,2,3)
Jules J. Berman, Ph.D., M.D. (2,3)
G. William Moore, M.D., Ph.D. (2,3,4)
Robert A. Yetter, Ph.D. (1,3)
From Research Service (1) and Laboratory Service (2), Department of Veterans Affairs Medical Center, Baltimore, Maryland; Department of Pathology, University of Maryland School of Medicine, Baltimore, Maryland (3); and Department of Pathology, The Johns Hopkins Medical Institutions, Baltimore, Maryland (4).
Address correspondence and reprint requests to: Dr. J. Seidman, Anatomic Pathology Laboratories, University of Maryland Medical Systems, 22 South Greene Street, Baltimore, MD 21201.
Running Title: Keratoacanthoma Flow Cytometry
Key words: Keratoacanthoma, squamous cell carcinoma, aneuploidy, flow cytometry, multiparameter analysis
Keratoacanthomas (KAs) are rapidly growing cutaneous lesions that frequently look much like well-differentiated squamous cell carcinomas (SCCs), but spontaneously regress. It is uncertain whether KA is a reactive hyperplastic lesion that mimics a neoplasm, or is a true (but defective) neoplasm that cannot sustain progressive growth. To address this question, we performed DNA flow cytometric analysis on fourteen KAs and ten cutaneous SCCs for comparison. By multiparameter DNA flow cytometry using forward scatter and orthogonal scatter, ten KAs and four SCCs had peridiploid DNA aneuploid populations (DNA indices of 1.03 to 1.14), and two SCCs had grossly aneuploid populations (DI 1.69, 2.33). Our data thus support aneuploidy in KAs. It is argued that KA is a true neoplasm.
Keratoacanthomas (KA) are rapidly growing cutaneous lesions that spontaneously regress. Morphologically, they frequently look much like well-differentiated squamous cell carcinomas. Is keratoacanthoma a reactive hyperplastic lesion that mimics a neoplasm, or is it a true (but defective) neoplasm that cannot sustain progressive growth? This is an important question that is at the heart of the study of incipient neoplasia.
In favor of a hyperplastic origin is its rapid origin (weeks), compared to a slow growth (years) exhibited by most neoplasms. In addition, although there are rare reports of neoplasms that regress, there are no other examples of neoplasms that regress with such regularity as do keratoacanthomas.
A neoplastic origin for keratoacanthoma is supported by its close morphologic similarity to well-differentiated squamous cell carcinoma, including infiltrating margins, cytologic atypia, and rare atypical mitotic figures.8 Occasionally, tumors with morphologic features of keratoacanthoma do not regress, but continue to grow and behave in a manner clinically indistinguishable from well-differentiated squamous cell carcinoma 21.
DNA ploidy analysis is an excellent tool for determining whether a lesion contains a population of cells with a somatic genetic abnormality.3,14 The presence of a clonal population of aneuploid cells from a lesion is usually considered evidence that the lesion is neoplastic 2. Although some reactive lesions have been shown to contain cells with aneuploidy, these cells are generally considered to represent a minor subpopulation that has sustained genetic injury and that are incapable of further cell replication.
Flow cytometry permits analysis of large numbers of cells from lesions as well as sampling of many lesions culled from a paraffin archive. In this study, KAs and cutaneous squamous cell carcinomas (SCC) were studied by flow cytometry to determine whether KA, like SCC, may contain a DNA aneuploid population.®PG¯
Paraffin blocks of all KAs and cutaneous SCCs diagnosed from May, 1988, through October, 1989, were retrieved from the surgical pathology division of Laboratory Services of the Baltimore Department of Veterans Affairs Medical Center. All tissues had been fixed with 10% neutral buffered formalin. The dimensions of the tumors were transcribed from he gross description of the surgical pathology reports, or directly measured from the H&E section of the tumor. All slides were reviewed to confirm the diagnoses. Diagnostic features of keratoacanthomas included a central keratin crater, large glassy cells at the edges and base of the lesion, and a distinct edge/collarette to the tumor at its epidermal borders.
Formalin-fixed, paraffin-embedded tissues were prepared for flow cytometric analysis as described by Hedley et al 11 and as modified by Hitchcock et al 12. Briefly, two to five 70-micron sections were deparaffinized with HistoSolv (Biochemical Sciences, Inc., Bridgeport, NJ), rehydrated with Flex (Richard Allan Medical Industries, Richland, MI) and digested for two hours with 0.2% pepsin (Sigma Chemical Co., St. Louis, MO). The pepsin, ribonuclease A, propidium iodide, and phosphate-buffered saline (PBS) solutions included 3% polyethylene glycol-8000 (Sigma) and the 0.1% Triton X-100 PBS solution contained 1 mg/ml bovine serum albumin (Sigma). Isolated nuclei were examined on an Epics Profile Flow Cytometer (Coulter Electronics, Hialeah, FL).
Forward scatter versus log side scatter plots were examined. In cases with two discrete cell populations, bitmap gating was used to generate a separate histogram for each subpopulation (fig. 1); the histogram for the total population was also examined.
Histograms were analyzed using Epis Profile software, version 2.0. In gated subpopulations, the mean channel number of the population with the greater amount of DNA was divided by that of the population with the lesser amount of DNA to obtain a DNA index (DI). Coefficients of variation [CV;(standard deviation/mean)x100] were obtained on the diploid peak of the ungated sample. Proliferative fraction was calculated as (S+G2/M)/(G0/G1+S+G2/M). A check of the flow cytometer was performed (as recommended by Givan 9) to insure a channel number of zero for unstained particles. No correction was required.
Fourteen KAs and ten SCCs were studied. The data are shown in Tables I and II and figs. 2 and 3. For KAs, ungated SPF (S-phase fractions) ranged from 3.2 to 12.4% (mean 5.9%), and PF (proliferative fractions) from 6.2 to 22.2% (mean 10.9%). Ten KAs showed two peridiploid populations with DIs ranging from 1.04 to 1.14 (mean 1.08). The population with the higher side scatter and forward scatter was always the more proliferative population (higher percentage of population in S, G2, or M phases) and, in all but one case, had a higher mean channel number. In the KAs with two populations, the more proliferative populations (population A) had SPF from 3.5 to 12.9% (mean 7.5%), and PF from 6.3 to 39.4% (mean 19.6%). The less proliferative populations (population B) had SPF from 1.5 to 3.5% (mean 2.3%) and PF from 2.3 to 5.6% (mean 4.1%). CVs of the diploid peaks ranged from 5.3 to 11.8%.
Histologic examination showed that four KAs had a high degree of morphologic atypia characterized by nuclear abnormalities, occasional atypical mitoses (fig. 4), and highly infiltrative borders (Cases 11-14). Among these four cases, three had separable (by gating) populations with high DIs (1.10 to 1.14), and one showed a single population.
Of the ten SCCs analyzed, two were grossly DNA aneuploid, with DNA indices of 1.69 and 2.33. Three others showed two peridiploid populations with DIs ranging from 1.03 to 1.07 (mean 1.05). As with KAs, the population with higher scatter was, in every case, the more proliferative population, and consistently had a higher mean channel number.
Ungated SPF for SCCs ranged form 1.9 to 7.5% (mean 4.3%), and PF ranged from 4.9 to 14.2% (mean 8.2%). These numbers exclude the two obviously DNA aneuploid cases (cases 5 and 9) which, because of the aneuploid peaks, do not yield reliable estimates of cell cycle phases. In the three cases with two peridiploid populations, the more proliferative populations had SPF from 3.6 to 13.4% (mean 7.3%), and PF from 10.4 to 22.2% (mean 14.4%). The less proliferative populations had SPF from 2.6 to 3.8% (mean 3.2%), and PF from 5.3 to 7.7% (mean 6.6%). CVs of the diploid peaks ranged from 3.8 to 14.2%, and were constantly lower in gated subpopulations than in ungated histograms.
No correlations were found between tumor size or DI and any cell cycle fraction in either SCC or KA, nor were significant differences found between the two tumors regarding gated or ungated cell cycle fractions, DI, or frequency of a second cell population (aneuploid or peridiploid).
The DNA ploidy of KAs has been examined in several previous reports.7,19,20,23 Newton et al 19 reported aneuploidy in 48% of KAs and 58% of SCCs, but these authors did not include histograms of their flow data for evaluation. Frentz et al 7 studied two KAs and three SCCs but used external controls to define a diploid peak. That is to say, they determined the expected location of the diploid peak by running a flow study of normal tissue removed from a separate block and superimposing the diploid location into the KA histogram. Similarly, Randall et al 20, in their study of 42 skin tumors, used an external diploid control. The practice of using external controls in flow cytometry has been employed by a few authors to characterize aneuploidy in tumor populations whose histograms produce a single peak. Unfortunately, it has been proved that external controls cannot be employed because the avidity of propidium iodide for nuclear material varies considerably from block to block secondary to variation in conditions of tissue storage and fixation. 14 This causes the normal diploid channel number to vary with different `standard' external controls. This problem is particularly vexing in the case of tumors with small variations from diploidy.
For our study, we adopted a technique for distinguishing aneuploid populations by gating multilparameter data available from DNA flow cytometric analysis. Populations of heterogeneous cell types can often be separated by variations in the direction and intensity of deflection of incident light. Flow cytometers that can isolate the data generated by these subpopulations permit a controlled comparison of the ploidy of these subpopulations. Non-neoplastic populations will have subpopulations with superimposable DNA peaks. Heterogeneous populations that combine neoplastic and non-neoplastic cells may show subpopulations with distinct DNA peaks.
Multiparameter flow cytometric analysis has been used by several investigators 1,4-6,10,13,17,22,24,25. Although it is not yet in widespread use, nor is it standardized, this technique has proved to be invaluable in selparating cell populations that are otherwise unseparable by DNA analysis alone. Forward light scatter with or without orthogonal light scatter have been used alone 17,24 and with DNA analysis 1,4-6,13,22,25 to separate cell populations, including near diploid cell populations 1,4-6,13,22 Analysis of complex data often obtained with multiparameter flow cytometry has also been addressed. 15,18 Separation of two peridiploid populations with different DNA content is proof of DNA aneuploidy. In contrast, aneuploidy can be determined only by karyotypic analysis. When the DNA content of two discrete peridiploid populations differs even slightly, the fact that thousands or tens of thousands of nuclei from individual lesions are usually examined virtually assures the statistical significance of small differences in DNA content. This also suggests the existence of separate populations.
In this study, the gated subpopulations of KAs and SCCs with higher mean peak channel number were also the subpopulations with the higest percentage of cells in S, G2, or M phase. This observation fits the expectation that the cells that are aneuploid (tumor cells) are more likely to be dividing that the normal (control) cells in the tumor and adjacent tissue. In addition, these hyperdiploid populations had higher orthogonal and forward scatter. Forward scatter is related to particle size, and orthogonal scatter to the number of particles or membranes with changes in refractive index in the path of incident light. Aneuploid cells with more DNA would be expected to be larger as well as have a more complex chromatin, thus resulting in more changes in refractive index.
The significance of a wide histogram peak is controversial. A wide peak (that is, a peak with a high CV) may represent either two populations of cells with slightly different quantities of DNA, or may simply be the result of poor sample preparation or poor fixation. We have separated two distinct cell populations that have similar but not identical amounts of DNA in 13 cases in this study. Because the channel numbers of these populations are so close, and thus the DIs so close to unity (most less than 1.1), one could argue that they represent different cell cycle phases of one population; the one with higher scatter would thus correspond to the more proliferative phase. However, McFadden et al 16 looked at 20 flow histograms with CVs greater than 5.51 and confirmed aneuploidy in all 20 cases by image analysis. Of nine cases with CV less than 4.41, image analysis showed diploidy. Our clear distinction of two populations in our 13 cases, in light of McFadden's findings, strongly suggests that these tumors are aneuploid. Furthermore since KAs are well differentiated lesions, one would not expect them to have a high DI. It is true that a DI of less than 1.1 or 1.2 is often questionable if the data are derived from DNA analysis alone. However, when gating shows separate populations with GAussian DNA distributions and similar CV's, DNA aneuploidy is the most likely explanation.
Thus, five parameters suggest two distinct populations: forward scatter, orthogonal scatter, fluorescence (DNA content), proliferative activity and CV. The first four parameters show differences between the two populations. CV consistently decreased with gating, thus suggesting that the ungated histograms contain a mixture of cells with differing DNA contents, yielding a wider CV.
Our data support aneuploidy in KAs. In addition, the aneuploidy in KAs is, by flow cytometry, indistinguishable from that seen in cutaneous squamous cell carcinomas (except in grossly aneuploid SCCs). This observations does not preclude the existence of distinct genetic lesions in KA and SCC.
Studies to determine the genetic lesions in KA using cytogenetic and molecular biological techniques would be useful. Molecular markers might determining whether KAs have molecular lesions (e.g. oncogenes) in common with SCCs. The fundamental question of wheter KAs are neoplastic, however, requires fundamental knowledge of the key lesions that are both necessary and sufficient for neoplasia (i.e. KAs may have some but not all of the genetic lesions occurring in SCC or may have alll of the lessions occurring in SCC in addition to genes that suppress that cause tumor regression.
The importance of understanding as much as we can about keratoacanthomas should not be minimized by the relatively insignificant clinical challenge posed by this neoplasm. On the contrary, it may prove crucially important to have a model, understood on the molecular level, of tumor regression.
1. Banner BF, Chacho MS, Roseman DL, Coon JS. Multiparameter flow cytometric analysis of colon polyps. Am J Clin Pathol 87:313-318, 1987.
2. Barlogie B. Abnormal cellular DNA content as a marker of neoplasia. Eur J Cancer Clin Oncol 20:1123-1125, 1984.
3. Coon JS, Landay AL, Weinstein RS. Advances in flow cytometry for diagnostic pathology. Lab Invest 57: 453-479, 1987.
4. Coon JS, Weinstein RS. Nuclear flight scatter as a "second parameter" in flow cytometry of archival tumor specimens. Cytometry 9(Suppl):36, 1988 (abstract)
5. Diamond LW, Braylon RC. Flow analysis of DNA contnet and cell size in non-Hodgkins lymphoma. Cancer Res 40:703-712, 1980.
6. Eriksen B, Miller DS, Murad TM, Lurain JR, Bauer KD. Dual-parameter flow cytometric analysis coupling the measurements of forward-angle light scatter and DNA content of archival ovarian carcinomas of low malignant potential. Anal Quant Cytol Histol 13:45-53, 1991.
7. Frentz G, Moller U, Larsen JK. DNA flow cotometry of human epidermal tumours: Intra- and intertumour variability in ploidy and proliferative characteristics. Virchows Archiv B 48:175-183, 1985.
8. Giltman LI. Tripolar mitosis in keratoacanthoma. Acta Dermatovener (Stockholm) 61:362-363, 1981.
9. Givan AL, Shenton BK, Carr TW. A correction required for calculation of DNA ratios in flow cytometric analysis of ploidy. Cytometry 9:271-274, 1988.
10. Hedley DW. Flow cytometry using paraffin-embedded tissue: five years on. Cytometry 10:229-241, 1989.
11. Hedley DW, Friedlander ML, Taylor IW, Rugg CA, Musgrove EA. Method for analysis of cellular DNA content of paraffin-embedded pathological material using flow cytometry. J Histochem Cytochem 31:1333-1335, 1983.
12. Hitchcock CL, Norris HJ, Khalifa MA, Wargotz ES. Flow cytometric analysis of granulosa tumors. Cancer 64:2127-2132, 1989.
13. Koss LG, Czerniak B, Herz F, Wersto RP. Flow cytometric measurements of DNA and other cell components in human tumors: a critical appraisal. Hum Pathol 20:528-548, 1989.
14. McFadden PW, Clowry LJ, Daehnert K, Hause LL, Koethe SM. Image analysis confirmation of DNA aneuploidy in flow cytometric DNA distributions having a wide coefficient of variation of the G0/G1 peak. Am J Clin Pathol 93:637-642, 1990.
15. Newton JA, Camplejohn RS, McGibbon DH. A flow cytometric study of the significance of DNA aneuploidy in cutaneous lesions. Brit J Dermatol 117:169-174, 1987.
16. Randall MD, Geisinger KR, Kute TE, Buss DH, Prichard RW. DNA content and proliferative index in cutaneous squamous cell carcinoma and keratoacanthoma. Am J Clin Pathol 93:259-262, 1990.
17. Schnur PL, Bozzo P. Metastasizing keratoacanthomas? The difficulties in differentiating keratoacanthomas from squamous cell carcinomas. Plast Reconstr Surg 62:258-262, 1978.
18. Stephenson TJ, Cotton DWK. Flow cytometric comparison of keratoacanthoma and squamous cell carcinoma. Brit J Dermatol 1988;118:582-583.
A: Schematic representation of light scatter distribution of a
keratoacanthoma with two peridiploid cell populations.
B: Actual gating of light scatter distribution (case 12).
In the light scatter distribution (left graph) the x-axis
represents forward scatter and the y-axis log orthogonal
scatter. The two relatively discrete collections of points
within the designated regions are separated, and a histogram
is generated for each. In the histograms, the fractions
bounded by the numbered bars are:
(1) G0/G1 fraction;
(2) proliferative fraction (S+G2/M);
(3) G2/M fraction.
Figure 2: Flow cytometric histograms of two keratoacanthomas.
A: Ungated version. B: Gated version (Case 12).
C: Ungated version. D: Gated version (Case 13).
Both gated versions represent the more proliferative populations.
Note the higher mean channel numbers (figure above peaks;
directly proportional to DNA content) and increased S and G2/M
fractions (larger second peak) in the gated versions.
Figure 3: Flow cytometric histograms of an aneuploid squamous cell
carcinoma (Case 5). The DNA index (ratio of the mean channel number
of the aneuploid peak to that of the diploid peak) is 1.69.
Figure 4: Tripolar mitosis in keratoacanthoma (Case 13; hematoxylin
and eosin x 500).
Ungated A B
# %S %PF %S %PF %S %PF DI CV(%) Size(cm)
1. 3.6 16.7 5.4 39.4 1.9 5.4 1.06 7.5 0.2
2. 6.4 9.6 12.9 20.2 3.4 4.6 1.04 8.4 2.5
3. 3.2 10.8 3.6 22.4 1.5 3.0 1.06 7.3 0.7
4. 6.9 11.4 7.2 13.9 2.6 5.1 1.06 7.6 0.5
5. 3.8 8.5 5.7 1.0
6. 4.1 6.2 3.6 9.4 1.6 2.3 1.07 7.4 0.4
7. 4.3 7.7 7.0 13.5 2.5 3.9 1.06 6.1 0.5
8. 8.4 10.8 5.4 0.3
9. 3.6 7.1 7.0 0.5
10. 10.9 13.9 7.5 12.6 1.7 3.4 1.10 5.4 1.9
*11. 4.1 8.4 7.4 1.0
*12. 6.0 12.4 11.8 27.7 3.0 4.8 1.10 7.7 0.6
*13. 12.4 22.2 12.2 30.2 3.5 5.6 1.10 5.3 4.0
*14. 4.3 7.4 3.5 6.3 1.6 2.5 1.14 11.8 1.4
mean 5.9 10.9 7.5 19.6 2.3 4.1 1.08 7.1 1.1
*Keratoacanthoma with atypical histologic features.
SQUAMOUS CELL CARCINOMA
Ungated A B
# %S %PF %S %PF %S %PF DI CV(%) Size(cm)
1. 5.2 10.1 5.0 10.6 3.8 7.7 1.04 7.7 0.3
2. 3.2 7.0 4.0 0.1
3. 1.9 4.9 6.3 0.5
4. 3.5 7.6 4.4 0.7
5. (1.69) 6.8 0.5
6. 4.3 6.8 3.8 0.9
7. 2.9 6.3 3.6 10.4 2.6 6.9 1.03 14.2 0.4
8. 5.8 9.0 13.4 22.2 3.2 5.3 1.07 11.0 0.7
9. (2.33) 8.9 0.4
10. 7.5 14.2 6.6 0.6
mean 4.3 8.2 7.3 14.4 3.2 6.6 1.05 7.4 0.5
Last modified: January 17, 2008