Search

Mikroskopsko kvantitativno vrednotenje analize FISH je zamudna, naporna in za napake dovzetna metoda, ki zahteva vsaj dva usposobljena preiskovalca. Pri velikem številu vzorcev bi bilo smiselno v vsakdanjo prakso uvesti avtomatsko štetje, kar je izvedljivo s pomočjo ustrezno programsko opremljenega motoriziranega fluorescenčnega mikroskopa. Namen našega dela je bila primerjava rezultatov analize FISH, ki jih dobimo z ročnim mikroskopskim pregledom, z rezultati, ki jih dobimo s pomočjo avtomatskega štetja. Določili smo pražne vrednosti za avtomatsko analizo za nekatere sonde in določili parametre kontrole kakovosti pri avtomatski analizi FISH. Preverili smo tudi ponovljivost avtomatskega vrednotenja in primerjali čas analize med ročnim in avtomatskim štetjem. V raziskavo smo vključili 85 preparatov, vzorci so pripadali 55-im bolnikom. Vzorce smo nacepili na gojišča, jih gojili 24–72 ur, nato smo celice izolirali. Vzorce smo nanesli na objektna stekla, jih označili z različnimi DNA-sondami in izvedli preiskavo FISH. Preparate smo vrednotili najprej ročno, nato pa še avtomatsko. Ugotovili smo, da lahko uporabimo le sonde, s katerimi ugotavljamo pomnožitev ali izgubo signalov in še te z določenimi omejitvami. Pri določenem deležu pozitivnih signalov moramo namreč s pomočjo programske opreme ročno prerazporediti celice glede na vzorec signalov, da bi lahko potrdili ali ovrgli prisotnost kromosomske spremembe. Avtomatsko štetje pravilno prepozna okoli 75–85 % celičnih jeder, medtem ko jih moramo okoli 15–25 % prerazporediti. Delež nerazporejenih oziroma izključenih celic je različen, odvisno od tega ali so prisotni dodani razredi razporeditve ali ne. Brez dodanih razredov jih je okoli 15 %, drugače pa precej več. Pri pozitivnih vzorcih med ročnim in avtomatskim načinom štetja ni statistično značilnih razlik (p > 0,05), korelacija med različnima načinoma štetja pa je večinoma zelo dobra do odlična. Translokacijskih in razcepnih sond, kjer nastane ali pa se razcepi zliti signal (fuzija), nismo uspeli validirati, saj avtomatskemu štetju težave povzročajo že negativni (normalni) preparati. Zaradi tega smo navedli tudi možne vzroke in možne rešitve. Ponovljivost avtomatskega vrednotenja, tako ponovljivost celotne analize kot tudi samega štetja je ustrezna, prav tako je s hitrostjo analize, ki je med ročnim vrednotenjem in avtomatskim štetjem primerljiva. V vsakdanjo prakso bi bilo torej smiselno uvesti avtomatsko štetje le za sonde, kjer se spremeni število signalov. The microscopic quantitative evaluation of the FISH analysis is a time-consuming, fatigue and error-prone method that requires at least two qualified investigators. For a large number of samples, it would be sensible to introduce automatic counting into the routine, which is possible using a motorized fluorescence microscope with appropriate software. The purpose of the study was to compare the results of the FISH analysis obtained by manual microscopic examination with the results obtained by automatic counting. For some DNA probes cut-off values for automatic analysis and quality control parameters for automatic FISH analysis were determined. The repeatability of automatic evaluation and comparison of the speed of analysis between manual and automatic counting were also observed. The study included 85 slides, with samples belonging to 55 patients. Samples were harvested after 24–72 hours cultivation. Samples were applied to object glasses, labeled with different DNA probes, then the FISH investigation was conducted. The slides were first evaluated manually and then automatically. It was observed that only probes which detected signals amplification or loss can be used and even those with certain limits. For some proportion of positive signals, software has to be used to manually reallocate cells according to the pattern of signals to confirm or reject the presence of a chromosomal rearrangement. Automatic counting correctly identifies about 75–85 % of cell nuclei, while we need to reallocate about 15–25 %. The proportion of unallocated (excluded) cells is different depending on the presence of the added allocation classes. Without added classes, the proportion is about 15 %, otherwise much more. There are no statistically significant differences between methods in positive samples (p > 0.05). Correlation between manual and automatic method of counting is mostly very good to excellent. It was not possible to validate the translocation and break-apart probes, because automatic counting already had difficulties caused by negative (normal) slides. Possible causes and possible solutions for this are listed in the study. Reproducibility of automatic evaluation, both the reproducibility of the entire analysis as well as the counting itself is appropriate. This is also true for the speed of analysis which is completely comparable between manual evaluation and automatic counting. Therefore, only for probes with numerical changes of signals it would be sensible to introduce automatic counting into the routine.