The shrinkage rate during preparation of histopathological specimens depends on specimen type and the preparation conditions, and varies from 14.7 to 47.3% in head and neck cancer [15-18]. Those previous studies reported on the shrinkage rate of mucosal longitudinal and/or transverse length. While surgical margins in head and neck cancer specimens easily shrink after formalin fixation, specimens were fixed by pins during formalin fixation in this study. No studies have clarified the width measured from the surface of mucosa to the deepest tumor site. This is the first report to evaluate the width shrinkage rate during preparation process of histopathological specimens fixed by pins. The pins fixed the specimen to reproduce an anatomical form, preventing shrinking of the mucous membrane, not compressed with weight of specimens in formalin. Here we showed that the shrinkage rate of tongue width was 10.3%. Therefore, c-DOI could be calculated according to the following formula; c-DOI = p-DOI × 100/89.7.
Of note, the shrinkage rate depends on the presence or absence of pins, formalin dipping time, paraffin infiltration process, room temperature, etc. Thus, the shrinkage rate may vary between facilities. To ensure consistency, in this study, all procedures were performed in constant conditions by a single pathologist.
There are many comparative studies between r-DOI and p-DOI using ultrasound [8, 19] or MRI [3, 7, 20, 21]. Several studies stated that CT [22, 23] is also useful for evaluation of r-DOI; however, CT cannot evaluate r-DOI when the contrast is low, superficial lesions cannot be detected, and metal artifacts often disturb images of tumor location. Therefore, we used ultrasound and MRI. Yesuratnam et al.  reported the correlation between tumor thickness on ultrasound and MRI with histopathologically determined TT of tongue carcinoma. Preoperative TT determined by US demonstrated higher correlation with pathological TT, compared with TT determined by MRI. However, there were several problems in this report , including evaluating histopathological TT instead of p-DOI, inflammation due to biopsy as MRI was taken after biopsy, and ignorance of the shrinkage of specimens. Our study excluded cases of exophytic tumor and remarkably ulcerous lesions because measurement of r-DOI in these tumor types is unreliable. r-DOI on MRI was divided into “before biopsy” and “after incision biopsy”, and considering the width shrinkage rate of specimens during preparation of histopathological specimens.
Ultrasound has advantages as being an easy method and can easily assess even superficial tumors because of a sufficiently high contrast. Yet, the measurement of exophytic tumor and remarkably ulcerous lesions is difficult, and it is hard to insert the probe in case of a posterior tumor of the tongue.
The advantage of the MRI is that the image is acquired in the tongue’s resting position and that there is no upper limit of the measured value. The disadvantages include difficult detection of superficial tumors, influence of metal artifacts, and movement of the tongue. We used MRI for cases with p-DOI > 3 mm, because 3 mm is an MRI cutoff value of detectable lesions. Similarly, Baba et al. reported that the cutoff value of p-DOI for detectable lesion on MRI was 4 mm . Therefore, the detection limit of p-DOI on the MRI is likely 3–4 mm.
Several studies have reported a significant relationship between r-DOI measured on MRI and p-DOI in the tongue SCC [3, 7, 20, 21]. To the best of our knowledge, this is the first study on MRI divided into “before biopsy” and “after incision biopsy”. Our prior study focusing on the period between 2006 and 2015 showed that r-DOI using MRI before biopsy best correlated with p-DOI . The present study expanded on this by including a larger patient number and considering the width shrinkage rate during the preparation process of histopathological specimen. Finally, similar results were obtained. It is desirable that the regression equation of c-DOI is y = x, which would indicate that c-DOI and r-DOI are equal. The regression equation of MRI before biopsy was y = 0.89 * x − 0.26, suggesting slight overestimation of c-DOI. This is probably due to reflection of stromal reaction around the tumor, such as lymphoplasmocytic infiltration. However, the coefficients of determination were very high (0.891), and r-DOI using MRI before biopsy was the most reliable. The regression equation of MRI after biopsy was y = 0.52 * x + 2.63, indicating severe overestimation of c-DOI because of inflammatory reaction of the tongue muscles caused by biopsy [4, 27]. Biopsy may lead to edema or hemorrhage and subsequent overestimation of tumor size and invasion depth in MRI [4, 24]. Especially in small tumors, if they are biopsied prior to imaging, the inflammation may affect the MRI interpretation. Further studies are necessary to determine how much the inflammatory reaction of biopsy spreads and how long an inflammatory reaction after the biopsy continues on MRI.
On the other hand, the regression equation for ultrasound was 1.12 * x + 0.21. The coefficient of determination was 0.664. This means slight underestimation of c-DOI and a wide distribution. These caused influencing to measure putting forward tongue, the period from the image to surgery was longest, and a border may not depict the boundary unclear such as high grade of histopathological pattern of invasion.
Our study had several limitations. First, the shrinkage rate was measured at the maximal width of surgical specimen, not of DOI of the tumor. However, measurement of DOI at the maximal cross-sectional area of the tumor before formalin fixation made making a slide at the same slice difficult. Second, the number of cases with MRI before biopsy was small (n = 18). We usually performed the biopsy at first presentation to operate earlier because MRI could not be taken immediately but there was a waiting time. Thus, patients were usually subjected to MRI after the biopsy in this study. Because the long waiting time until operation would likely influence to treatment results, an early MRI and operation is desirable. Although it is difficult to improve the waiting time until the operation in many facilities, a higher number of cases in future studies would be required.
In conclusion, the specimen shrinkage rate during the histopathological specimen preparation process was 10.3%. After correcting p-DOI accordingly, r-DOI using MRI before biopsy most strongly correlated with c-DOI.