Making Malign-Benign Separation of Ovarian Masses Using Logistic Regression Model

Öz

Aim: Aur aim was to fınd out the power of the joint use of pelvic examination, tumor markers and ultrasonoraphy for dedecting malignancy of ovarian masses and searching out the ovarian masses’ malignancy probability by applying this parameters to Logistic regression model.Material and Methods: Ages, complaints, menopausal statuses, pelvic examination and ultrasonography findings, CA-125 values, pathology reports of 154 patients operated with the diagnose of adnexial mass in our clinic were examined retropectively. Ultrasonographic findings of ovarian masses scored according to criterions identified by Sassone at all. 2 point for irregular mass surface, solid mass and presence of abdominal ascite are given in Our physical examination scoring system. Sensitivity and specificity of all parameters were determined. Probability of malignancy are calculated by Logistic regression model.Results: There were 55 malign 35.7% ,99 benign 64.3% , totally 154 women and 191 masses in our study. Sensitity and specificity of our pelvic examination scoring system were 92% and 94% respectively. Sensitivity and specificity of Ca-125 in discrimination of malign-benign adnexial masses were 77% and 73% respectively. When adnexial masses with malign evaluation of any two of three parameters Ca- 125 value, pelvic examination and ultrasonography were evaluated as malignant; sensitivity, specificity and overall accuracy rate were calculated as 93.33%, 95.24% and 94.44% respectively.Conclusion: Diagnostic models like Logistic regression models are formations as monotype, independent from operator’s experience, enabling to test accuracy of diagnose. This models by searching out the ovarian masses’ malignancy probability give way to correct decision of operation type, side also suitable operatör and to inform the patient before definitive diagnose.

Anahtar Kelimeler

• Logistic regression model,ovarian mass,pelvic examination,ultrasonography,CA-125

Lojistik Regresyon Modeli Kullanılarak Ovariyan Kitlelerin Malign-Benign Ayırımının Yapılması

Öz

Amaç: Amacımız pelvik muayene, tümör belirteçleri, ultrasonografi beraber kullanıldığında ovariyan kitlelerde malignansiyi tespit edebilme gücünü ortaya koymak ve bu parametreleri lojistik modele uygulayarak adneksiyal kitlesi olan bir hastadaki malignansi olasılığını tespit etmekti.Gereç ve Yöntemler: Kliniğimizde adneksiyal kitle tanısı ile opere edilen 154 hastanın patoloji kayıtları, yaşı, şikâyeti, menapoz durumu, pelvik muayene buğuları, CA-125 değerleri ve ultrasonografi bulguları retrospektif olarak incelendi. Ovariyan kitlelerin ultrasonografi bulguları Sassone ve arkadaşları tarafından tanımlanan kriterlere göre puanlandırıldı. Geliştirdiğimiz pelvik muayene skorlama sisteminde pelvik muayenede kitle yüzeyinin irregüler ve kitlenin solit oluşu ve batında asit varlığına 2 değer; kitlenin bilateral ve fikse oluşuna 1 değer; kitlenin kistik, düzgün yüzeyli, mobil, unilateral oluşuna veya batında asit olmayışına 0 değer verildi. Tüm parametrelerin sensitivite ve spesifite değerleri tespit edildi. Lojistik regresyon modeli ile malignansi olasılıkları hesaplandı.Bulgular: Çalışmamız 55 i malign %35.7 ,99 u benign %64.3 toplam 154 kadın hasta, 191 kitle üzerinde yapıldı. Geliştirmiş olduğumuz pelvik muayene skorlama sisteminde sensitivite %92, spesifite %94 olarak hesaplandı. CA-125’in malign-benign ayırımındaki sensitivitesi %77, spesifitesi %73 olarak tespit edildi. CA 125, pelvik muayene ve ultrasonografi değerlendirmelerinden herhangi ikisinde malign gruba düşen hastaları malign olarak değerlendirdiğimizde elde ettiğimiz sensitivite %93.33, spesifite %95.24, genel doğruluk oranı %94.44 bulundu.Sonuç: Lojistik regresyon modelleri gibi tanısal modeller adneksiyal kitlelerin değerlendirilmesinde tek tip, operatörün deneyimine bağlı olmayan, tanının doğruluğunun test edilebilmesine olanak sağlayan oluşumlardır. Bu modeller malignansi olasılığını tespit edebilmeye olanak sağlayarak operasyon tipine, yerine, hatta operatöre karar verebilmeye ve hastanın kesin tanı öncesi bilgilendirilmesine olanak sağlar.

Anahtar Kelimeler

• Logistic regresyon modeli,ovarian kitle,pelvic muayene,ultrasonografi,CA-125

Kaynakça

1. GLOBOCAN 2012: estimated cancer mortality, incidence and prevalence worldwide in 2012. http://globocan.iarc.fr/Pages/fact_sheets_populati- on.aspx.
2. Turgay Atasu. Jinekolojik Onkoloji. İkinci basım. 1999.
3. Lynda DR, Laıla IM, Sharon MS. Pelvik cxamination, tumor marker level and gray-scale and dopplcr sonography in the prediction of pelvic cancer. Obstet Gynecol 1997;89:493-500.
4. Curtın JP. Management of the Adnexal Mass 1994; 55:42-46.
5. Sassone AM, Tmıor-Tritsch IE. Transvagınal sonographic characterization of ovariyan disease:Evaluation of a new scoring system to predict ovari- yan malignancy. Obstet Gynecol 1991;78:70.
6. Roman LD, Muderspach LI, Stein SM, Narin SL, Groshen S, Morrow CP. Pelvic Examination, Tumor Marker Level and Gray-Scale and Doppler So- nograplıy in the Predictio of Pelvic Cancer. Obstet Gynecol 1997; 89:4.
7. Finkler NJ, Benacerraf B, Lavin PT, Wojciechowski C, Knapp RC. Compa- rison of serum CA 125, clinical impression ,and ultrasound in the preope- rative evaluation of ovariyan massess.Obsıet Gynecol 1998;72:659-64.
8. Strigini FAL, Gadducci A, Del Bravo B, Ferdeghini M, Genazzani AR. Diffe- rential diagnosis of adnexial masses with transvaginal sonography, color flow imaging and serum CA 125 assay in pre and postmenopausal wo- man. Gynecol Oncol 1996; 61:68-72.

9. Vasilev SA,Schaerth JB, Campcau J, Morrow CP. Serum CA 125 lcvcls in preoperative evaluation of pelvic masses.Obstet Gynecol 1998; 71:751- 61.
10. Di-Xia C, Schwartz PE, Xinguo L, Zhaaan Y. Evaluation of CA 125 levels in differentiating malîgnant tumors from benign tumors in patients with pelvic masses. Obstet Gynecol 1988; 72:23-7.
11. Shutter EMJ,Sohn C, Kristen P, Möbus V. Estimation Of probability of Malignancy Using a Logistic Model Combining Physical Examination, Ultrasound, Serum CA-125 and Serum CA72-4 in Postmenapausal wo- men with a Pelvic Mass: An International Multi center Study.Gynecologıc Oncology1998; 69:56-63.
12. Van Nagel JR, Higgins RV, Donaldson ES. Transvaginal sonography as a screening method lor ovariyan cancer : A report of the first 1000 caşes şcreened. Cancer 1990; 65:573-577.
13. Van NagclI.J.R..De Priest .P.D..Puls ,L.E.,et al.Ovariyan cancer screening in asympıomatic postmenopausal women by transvaginal sonography .Cancer 68,485-462 ,1991.
14. Shutter EMJ, Kenemans P, S Christof, Kristen P. Diagnostic Value of Pelvic Examination Ultrasound and Serum CA 125 in Postmenopausal Women with a Pelvic Mass 1994; 74:4.
15. Timmerman D, Bourne TH, Ataitor, Collins WP, Verrelst H. Comparison of methods tor preoperative discrimination between malignant and benign adnexal masses: The development of a new logistic regression model. A. J. Obstet Gynecol 1999;181:1.
16. Aslam N, Banerjee S, Carr JV, Savvas M, Hooper R, Jurkovıc D. Pros- pective Evaluation of Logistic Regression Models tor the Diagnosis of Ovariyan Cancer. Obstet and Gynecol 2000;96:1.
17. Tailor A, Jurkovic D, Bourne TH, Collins WP, Campbell S. Sonographic prediction of malignancy in adnexal masses using multi variate logistic regression analysis . Ultrasound Obstet Gynecol 1994; 10:41-7.
18. Alcazar JL, Jurado M. Using a logistic model to predict malignancy of adnexal masses hased on menopausal status, ultrasound morphology and color doppler findings. Gynecol Oncol 1998; 69:146-50.