Radiol Oncol 1998; 32(1): 65-108
Longitudinal study of malignancy associated changes in progressive cervical dysplasia
Margareta Flezar1, Jaka Lavrencak1, Mario Zganec2, and Marija Us - Krasovec1
1Institute of Oncology, 2Faculty of Electrical Engineering, Ljubljana, Slovenia
Eight of 29 patients with progressive CIN were followed for 2 to 10 years. Their consecutive Pap smears were destained and stained according to Feulgen thionin method. Cyto-Savant™ high resolution image cytometer (Oncometrics Technol. Corp., Vancouver, Canada) was used far image acquisition and analysis. Average values of nuclear texture features and their probability distributions far consecutive Pap smears from each patient were calculated. Three out of 5 discriminant ^AC, highDNAamount, highD-NAcomp and highDNAarea, were found to increase as a function of time in 5 out of 8 patients. A preliminary analysis which was performed on non-standardized archival material demonstrated a monotonous increase of discrete texture features as a function of time in patients with progressive CIN.
Key words: cervix dysplasia; image cytometry; cell nucleus
Introduction
The goal of an efficient screening of uterine cervical smears is the eradication of invasive cancer of the cervical squamous epithelium. This could be achieved by an early detection and subsequent treatment of precancerous lesions. However, the debatable issue remains the treatment of moderate dysplasias (CIN 2). The follow up of the patients with moderate dysplasias that were not treated showed that some of these lesions regressed to mild dysplastic lesions or even regressed back to normal squameous epithelium, while some progressed to severe dysplasia, carcino-
Correspondence to: Margareta Flezar, MD, MSc, Institute of Oncology, Dept. Cytopathology, Zaloska 2, 1105 Ljubljana. Tel: +386 61 323 063 ext. 4946; Fax:+386 61 1314 180; E-mail: mflezar@onko-i.si
ma in situ and invasive carcinoma. At the time of diagnosis of moderate dysplasia one cannot predict whether the lesion will progress and whether the treatment of the lesion should be administered. Additional diagnostic tools would therefore be helpful to establish the biological potential of moderate dysplasias.
Some authors claim that the analysis of cells and - or nuclei by means of image cytometers can yield data about the progression or regression of the lesions.1,2,3 Their hypothesis is based on measurements of chromatin structure and organization (nuclear texture features). Moderate dysplastic lesions that will progress to cancer differ from the ones that will regress mainly in nuclear texture features. The combination of nuclear texture features discriminating
66
Flezar M et al.
between the two lesions that have a different biological potential is described as MAC (malignancy associated changes) or in some other studies as kariometric factors.1,2,4
In our previous study we found significant differences between progressive and regressive moderate dysplasias (CIN 2) by image cytometric analysis in five nuclear texture features which could be explained as MAC.5 In the present study our aim was to determine the time of onset of MAC which would be characteristic for progressive moderate dysplasias. The occurance of MAC was studied on normal intermediate cells present in the cervical smears obtained by the routine screening program.
Materials and methods
From the files of 3 different laboratories, 29 patients with progressive and 21 patients with regressive moderate dysplasias (CIN 2) were found. However, only 8 patients (ages 22 to 42 years) were followed up with at least three consecutive cervical smears before the diagnosis of progression to CIN 3 (severe dysplasia or carcinoma in situ) was established (Table 1). Their follow-up ranged from 2 to 10 years.
The cervical smears that were originally Papanicolaou stained were destained in acid alcohol and nuclear DNA was stained ste-chiometrically according to modified Feulgen
Table 1. Patient's diagnosis, age at the time of diagnosis and the length of follow-up in years
Patient	Diagnosis	Age at diagnosis	Follow-up
		(years)	(years)
1	CIN 3	23	2
2	CIN 3	33	6
3	CIN 3	29	9
4	CIN 3	29	7
5	CIN 3	23	6
6	CIN 3	42	10
, 7	CIN 3	25	6
8	CIN 3	22	3
method by thionin. Image cytometric analysis was performed by an automated high resolution image cytometer Cyto-Savant™ (Oncometrics Technol. Corp., Vancouver, Canada).6
From the several hundred nuclei acquired automatically, we selected an average of 76 ranging from 15 to 177, well preserved nuclei of normal (non-diagnostic) intermediate squamous cells per cervical smear. Over 100 nuclear features and most importantly nuclear texture features were calculated for each nucleus scanned, detailed description and formulas of nuclear features are described elsewhere.7 Probability distributions were calculated for each nuclear texture feature, however, longitudinal observation of MAC appearance was performed only for the nuclear texture features found to have discriminative power between progressive and regressive moderate dysplasias in our previous study (Table 2). These features belong to the group of discrete nuclear texture features and reflect subtle differences in condensation between the neighbouring chromatin particles.
Results
The longitudinal observation of discriminating nuclear texture features showed a monotonous increase of values of 3 out of 5 nuclear texture features as a function of time in progressive moderate dysplasias in 5 patients.
The values of "highDNAcomp" increased through the time, that is from the first smear, diagnosed as moderate dysplasia, to the last one diagnosed as severe dysplasia or carcino-rna (Figure 1). In five patients (patients 2, 5, 6, 7, 8) we observed a slight decrease, no increase, or slight increase of value between the first and the second smear. However, a substantial increase was found between their second and the third (last) smear. In two patients (patients 3, 4) the values of "highD-
Longitudinal study of malignancy associated changes in progressive cervical dysplasia
67
0.2 0
Figur'e 1. The dynamics of "highDNAcomp" in consecutive smears of patients with progressive dysplasias.
Table 2. Nuclear texture features, which discriminated between progressive and regressive dysplasias
Discriminative nuclear texture features
HighDNAarea: Fraction of the total nuclear area occupied by high condensed chromatin HighDNAamount: Ratio of JOD* of high condensed chromatin (JOD* high) to JOD* HighDNAcomp**: Shape of high condensed chromatin
* Jntegrated Optical Density ** highDNAcompactness
NAcomp" remained basically unchanged throughout the whole period, even though there was a slight increase between their first and the second smear. One patient (patient
1)	showed exactly a reverse pattern of nuclear feature changes, that is an overall decrease in the value of this feature throughout the period.
A very similar pattern in changes of values was found for "highDNAamount" (Figure
2).	Again, five patients (patients 2, 5, 6, 7, 8) showed slight or no increase or even some decrease between the first and the second smear, and significant increase between their second and the last smear. The same two patients (patients 3, 4) presented with relatively unchanged values of this feature throughout the period. One patient (patient 1), the same as above, showed significant decrease in "hiDNAamount" values from her first to the last smear.
In addition, the longitudinal observation of "highDNAarea" showed a similar pattern of changes to the other two features in the same five patients (patients 2, 5, 6, 7, 8) (Figure 3). For this feature the values for five patients were practically unchanged between the first and the second smear. Afterwards, the values increased substantially. Two patients (patients 3, 4) had the constant values of this feature through the time, and, again, in one patient (patient 1) the values decreased from her first to the last smear.
Other features that were discriminative in the previous study did not show a convincing pattern of increase in the majority of patients.
—	Patient 1
—	Patient 2
-	- - Patient 3
-	- Patient 4
—	Pafient 5
—	Patient 6
-	- • Patient 7
-	- Patient 8
2
1	2	3
Figur'e 2. The dynamics of "highDNAamount" in consecutive smears of patients with progressive dysplasias.
0.03
1	2	3
Fijgure 3. The dynamics of "highDNAarea" in consecutive smears of patients with progressive dysplasias.
3
68
Flezar Metal.
Discussion
Population screening of cervical smears for an early detection of preneoplastic lesions of uterine cervix is becoming an important issue in health care programs in different countries all over the world. Some countries have already established efficient screening programs that cover high percentage of women, while others are still in the process of introducing a widespread screening. A successful screening program detects more women with early preneoplastic changes, such as mild and moderate dysplasias of cervical epithelium.8 By light microscopy it is not possible to assess the biologic behaviour of cervical dysplasias. To avoid the overtreat-ment of women harbourning dysplasias with nonprogresive course, they should possibly be defined as such by an additional diagnostic analysis.
Image cytometry was used to measure the chromatin structure and organization in cervical neoplastic lesions objectively. Initially, the nuclei of diagnostic cells in dysplastic lesions were analyzed and cytometric differences between the nuclei of different grades of dysplasias were reported.9'10 These studies were mostly performed as a part of projects that would introduce automated image cyto-metric screening of cervical smears, and nuclear texture features were used to correct classification of the cells into normal and dysplastic ones.
Besides, the subvisual chromatin changes were also studied in normal intermediate cells, which were also present in the cervical smears together with dysplastic cells.11, 12, 13, 14, 15 Subtle chromatin changes were first detected by means of light microscopy examination in normal cells of patients with a malignant disease and were described as malignancy associated changes (MAC).16 With the development of high resolution image cytometers the MACs were objectively measured and they were found not only in
connection with different malignant tumors, but also in normal cells of dysplastic lesions, hence also in dysplasias of cervical epithelium. 1,2,11,12,13,14,15
Subsequently, the idea of MAC was introduced in the studies of progression or regression of cervical dysplastic lesions. Some authors claim that chromatin structure of progressive dysplastic lesion differs from the structure of regressive lesions.1,2,3 In our previous study we were also able to discriminate between the two types of lesions by using certain nuclear texture features.5 However, the dynamics of subvisual chromatin structure changes as a function of time has not been studied yet. In our small group of patients, followed-up for 2 to 10 years, the increase of discriminative nuclear texture features in subsequent smears was found in some patients in normal intermediate cells. Since the discriminative texture features from the previous study reflect the condensed portion of chromatin particles in the nucleus, the results of this study suggest a condensation of chromatin with the progression of the lesion. A larger data set with more patients involved in the study should be made to confirm our observations. Also, a larger and more consistent increase of the feature values could be expected, if there were no variations in the initial preparation (the effect of fixation and original staining), which were found to cause differences in nuclear texture features.17 Besides, nuclear texture features are also influenced by the age of the patients, vaginal inflammations and hormonal status.13,18
We expect that more extensive image cyto-metric studies on different grades of cervical dysplasias will define which patients would progress to more severe lesions and which lesions will regress. If several subsequent smears of the same patient were obtained through a more efficient screening program, we could retrospectively learn more precisely, when the exact progression begins. Accord-
Longitudinal study of malignancy associated changes in progressive cervical dysplasia	69
ingly, most appropriate treatment would be planned and most of all the aggressive overtreatment of patients, still in childbear-ing age, would be avoided.
Acknowledgment
This work was partly supported by the Slovenian Ministry of Science and Technology, grant No. J3-7956.
References
1.	Palcic B, MacAulay C. Malignancy associated changes. Can they be employed clinically? In: Weid GL, Bartels PH, Rosenthal D, Schenk U, ed. Compendium on the computerized cytology and histology laboratory. Chicago; Tutorials of cytology, 1994; 157-65.
2.	Palcic B. Nuclear texture: Can it be used as a surrogate endpoint marker? J Cell Biochem 1994; Suppl 19: 40-6.
3.	Hanselaar A, MacAulay C, Palcic B, Garner D, LeRiche J. Discrimination between progressive and regressive cervical intraepithelial neoplasia (CIN) by DNA-cytometry. Analyt Cell Pathol 1992; 4: 165, Abstract 134.
4.	Bibbo M, Montag AG, Lerma-Puertas E, Dytch HE, Leelakusolvong S, Bartels PH. Karyometric marker features adjacent to invasive carcinoma. Analyt Quant CytolHistol 1989; 11: 281-2.
5.	KodricT, But-Cigler N, Vrh- Jermancic J, Lavren-cak J, Zganec M, Us - Krasovec M. Nuclear texture features of normal intermediate cells in progressive and regressive/persistent cervical intraepithelial lesions. Acta Cytol 1997; 41: 1210, Abstract.
6.	Jaggi B, Poon S, Ponitifex B, Fengler J, Marquis J, Palcic B. A quantitative microscope for image cytometry. Proc SPIE 1991; 14A8: 89-97.
7.	Doudkine A, MacAulay C, Poulin N, Palcic B.
Nuclear texture measurements in image cytometry. Pathalogica 1995; 87: 286-99.
8.	Koss L. The Papanicolaou test for cervical cancer detection. A triumph and a tragedy. JAMA. 1989; 261: 737-43.
9.	Wied GL, Bartels PH, Dytch HE, Pishotta FT, Yamauchi K, Bibbo M. Diagnostic marker features in dysplastic cells from the uterine cervix. Acta Cytol 1982; 26: 475-83.
10.	Hanselaar A. DNA cytometry of cervical intraepithelial neoplasia. PhD Thesis, Katholieke Univer-siteit Nijmegen, October 1990.
11.	Bibbo M, Bartels PH, Sychra JJ, Wied GL. Chro-matin appearance in intermediate cells from patients from uterine cancer. Acta Cytol 1981; 25: 23-8.
12.	Bartels PH, Bibbo M, Dytch HE, Pishotta FT, Yamauchi K, Wied GL. Diagnostic marker displays for intermediate cells from uterine cervix. Acta Cytol 1982; 26: 29-34.
13.	Haroske G, Bergander St, König R. Frequency and diagnostic reliability of subvisual morphologic markers for malignancy in the cervical epithelium. Areh Geschwulstforsch 1988; 58: 159-68.
14.	Haroske G, Kunze KD, Dimmer V, Meyer W. Some remarks on preinvasive cervicalneoplasia -an image analysis study. Arch Geschwulstforsch 1985; 55: 245-52.
15.	Haroske G, Bergander St, König R, Meyer W. Application of malignancy-associated changes of cervical epithelium in a hierarchic classification concept. Anal Cell Pathol 1990; 2:189-98.
16.	Nieburgs HE, Goldberg AF, Bertini B, Silagi J, Pacheco B, Reisman H. Malignancy associated changes (MAC) in blood and Bone marrow cells of patients with malignant tumors. Acta Cytol 1967; 11: 415-23.
17.	Flezar M, Us - Krasovec M. Differences in nuclear texture features caused by different initial fixation and duration of hydrolysis. Anal Cell Pathol 1997; 13: 60, Abstract 16.
18.	Kwikkel HJ, Boon ME, ven Rijswijk MMM. Masking effect of hormonal contraceptives on discriminative features of visually normal intermediate cells in positive and negative cervical smears. Anal Quant Cytol Histol 1986; 8: 227-32.