Presented at the Society of Surgical Oncology 64th Annual Cancer Symposium; March 2-5, 2011; San Antonio, TX.
Margin status is a predictor of outcome for patients with liver malignancies, although what constitutes a negative margin is controversial. Traditionally, the completeness of resection is estimated by surgical histopathology of the resected specimen margin, despite the in situ margin being potentially more important. The true margin is often altered by parenchymal transection techniques. The authors propose that cytologic assessment of the in situ margin is more specific for determining the true margin.
A total of 84 patients with primary or metastatic liver tumors who were undergoing surgical resection were enrolled in this prospective Institutional Review Board-approved study. Specimen and in situ (patient) margins were assessed using a “scrape preparation” cytologic technique and compared with traditional surgical histopathology. Patients were followed for assessment of local disease recurrence.
Follow-up data were complete for 64 patients for a median of 37 months (range, 12 months-56 months). Twenty patients were excluded because of perioperative death (6 patients; 7%) or a follow-up of < 12 months. Seven patients (12.2%) had positive histopathologic specimen margins, but only 1 was found to be positive by cytology (1.8%). No in situ cytologically positive margins were identified along the cut edge of the liver remnant. The rate of intra- or extrahepatic recurrences was 56.7%, whereas the local recurrence rate was 1.8%. One patient with local recurrence demonstrated simultaneous intra- and extrahepatic disease recurrences and had negative margins by all methods of evaluation.
The liver is a frequent site of primary or secondary involvement with cancer. Hepatocellular carcinoma is the most common primary malignancy of the liver and one of the most common cancers worldwide, affecting over 1 million people annually.1 In addition to being a frequent site of primary tumors, the liver is also the most commonly encountered site of metastasis for many malignancies, particularly colorectal carcinoma.2 Surgical resection is the mainstay of curative therapy for patients with hepatic malignancies.3, 4 However, complete resection is often not possible, owing to the amount of tumor burden, tumor location, or underlying synthetic function precluding the removal of substantial liver parenchyma. The key to successful liver surgery is the complete removal of all known disease, including negative microscopic margins, while leaving adequate reserve for normal hepatic function. Nevertheless, the most important factor influencing survival in patients undergoing hepatectomy for primary and metastatic malignancies is histologically negative margins of resection, regardless of the width of the margin.3-5 However, what constitutes a negative margin remains to be determined.
Assessment of the margin of resection in liver surgery is often difficult and may be inaccurate, particularly when margins measure < 1 cm, given the various techniques of parenchymal dissection. Normal hepatic parenchyma is often crushed, torn, or aspirated from the specimen as vessels are dissected and divided. During major hepatectomy, the surgeon will often accept a close margin when the patient is at risk of not having adequate postresection liver volume for recovery. A close margin in combination with parenchymal destruction may result in a margin being falsely reported as positive, resulting in unnecessary, and potentially harmful, additional resection to obtain a negative margin and/or adjuvant therapy. Often the ability to achieve a negative margin reflects tumor biology rather than inadequate surgery. Conversely, the surgeon's clinical assessment of the margin in the operating room may be falsely negative when one is convinced that the true margin has been removed by aspiration devices or distorted by parenchymal dissection. The true margin, and arguably the only important margin, is the one remaining in the patient, not the one sent to pathology.
We assessed the margin of resection in situ by scraping cells from the cut surface of the liver once the specimen was removed and subjected these cells to cytopathologic evaluation. This assessment was then compared with the standard histologic method of margin assessment. This is similar to the touch preparation or imprint cytology method often used for the assessment of surgical margins in patients with breast cancer.6-8 However, we used this technique on the cut edge of the liver remaining in the patient. Finally, patients were followed along their normal postoperative course to assess for patterns of disease recurrence. Our hypothesis in undertaking the current study was that cytopathologic assessment of the in situ margin of resection would more accurately represent the true margin status and predict local disease recurrence.
MATERIALS AND METHODS
Patients were accrued from 2005 through 2008 according to an Institutional Review Board (IRB)-approved protocol, and all hepatic resections were performed by 2 surgeons at a single institution. Informed consent was obtained from all patients. Liver resection was undertaken using stapling or monopolar cautery devices according to the surgeon's preference, with the majority of transections being performed with vascular staple loads. All operations were undertaken with curative intent. Once the specimen was removed and all bleeding was controlled, 2 sterile glass slides were brought up into the field. All liver parenchyma near where the tumor was located was gently scraped twice. A standard smear of the scraped tissue was created using the second slide, thus creating duplicate slides for evaluation. This slide pair represented the in situ margin (obtained from within the patient). On the back operating table, the resected liver margin was also scraped and a smear created using a similar technique (specimen margin). Finally, the tumor closest to the margin of resection was bivalved by the surgeon and the tumor itself was scraped and a third set of smears created (tumor). This served as a positive control to verify that tumor cells could be identified for each specimen. All specimen collection was performed by 1 of 3 IRB-approved personnel.
Per protocol, if multiple resections were undertaken in 1 patient, each specimen was considered individually if they were spaced > 2 cm apart and were from separate anatomic segments of the liver so that local recurrence could be detected radiographically. For example, 2 separate lesions in the right and left lobes of the liver counted as 2 separate entries for statistical purposes. Two lesions located in the most medial and most lateral portions of the right or left lobes were also considered separately if they required separate resections. However, 2 lesions within 1 resection specimen or 2 lesions within the same segment of the liver but removed separately were not counted as separate lesions. In the latter case, the lesion with the closest macroscopic margin as determined by the operating surgeon was included in the study.
In total, 6 smears (2 from the in situ margin, 2 from the specimen margin, and 2 from the tumor) were created and submitted for blinded cytopathologic review. These were allowed to air dry during surgery. One of each pair of slides was stained with Romanowsky (Diff-Quik) or Papanicolaou stains using standard cytology protocols (Figs. 1-3). All slides were randomly labeled and read by a single cytopathologist (P.E.W.) who was blinded to any patient information, including histology. The surgical specimen was handled according to routine histopathology protocols. Histologic margin status for the submitted surgical specimen was determined by a single pathologist (W.L.F.) who was blinded to the patient information, operation undertaken, tumor type, and cytopathologic margin status. Final cytopathologic and histologic margin statuses were not available to the surgeon during the operation.
The primary objective of the current report was to determine the accuracy of using cytopathologic techniques to assess the margin of resection after hepatectomy using the standard histologic margin status of the submitted specimen as the referent standard. We next sought to determine whether the specimen's histologic and cytologic margin statuses were correlated with the in situ margin status. For these initial comparisons, all 84 specimens were used.
As a secondary objective, we sought to determine the impact of cytologic in situ margin status on the local recurrence rate. Because the majority of local recurrences occur within 1 year of surgery, 64 patients with a minimum of 12 months of follow-up (median, 37.1 months) were included. Imaging follow-up was available for all patients. Local recurrence was considered when radiographic evidence of recurrent tumor was observed along the line of previous resection by computed tomography (CT) or magnetic resonance imaging. Positron emission tomography (PET) scans, when obtained during the usual course of care, were used to identify local recurrence only when images were fused with CT images (CT/PET scans). In other words, PET scans were not used as the sole determinants of local recurrence. No imaging studies were obtained outside the standard routine postoperative follow-up. Typically, patients underwent imaging within 3 months of surgery and/or before the initiation of further therapy. All additional imaging was obtained as indicated at the discretion of the treating medical/surgical oncologist. All patients were treated exactly the same with regard to postoperative follow-up and adjuvant therapy, regardless of whether they had a positive or negative histologic margin.
First, the sensitivity and specificity of confirming the presence of malignancy in the positive control specimens by cytopathology were determined using surgical pathology (ie, histology) as the referent standard. The specific tumor type was not the focus of cytopathology samples; rather, it was the presence or absence of malignant cells. Second, the sensitivity and specificity of the cytopathologic specimen margins were determined using histology as the referent standard. Finally, we used cytopathology to evaluate the in situ margin after liver resection.
Among patients, there was an evenly distribution of men and women (42 of each), with an average age of 59.4 years. Major hepatectomy (ie, ≥ 3 combined sectors) was undertaken in 53 patients (62.4%), with only 17.9 % of patients requiring an intraoperative blood transfusion. The majority of hepatectomies were undertaken for metastatic adenocarcinoma (62 patients; 73%), with colorectal adenocarcinoma being the most common (Table 1).
A total of 85 specimens were obtained from 84 patients. First, the ability to identify diagnostic tumor cells by scrape preparation cytopathology and surgical pathology was confirmed (Figs. 1-3). There were 84 true-positive specimens (positive on both surgical pathology and cytopathology) and 1 true-negative specimen (negative on both surgical pathology and cytopathology) within the group (Table 2). In 7 tumor samples, metastatic colorectal cancer that had been treated with preoperative chemotherapy demonstrated necrotic debris by cytopathology but viable tumor cells on standard histologic assessment of the entire tumor. These were considered as true-positive results because they were clearly distinguishable from surrounding hepatocytes.
Table 2. Descriptive Statistics of Cytopathology Using Histopathology as the Referent Standard
Abbreviations: NPV, negative predictive value; PPV, positive predictive value; SE, standard error.
14% (SE: 4.5%)
90% (SE: 3.9%)
Fifty-nine specimens were able to be evaluated for in situ and specimen margin status. Compared with the tumor identification specimens, there were 7 (11.9%) positive specimen margins on histology, but only 1 was positive on cytology (1.7% with a true-positive finding). These 7 samples are listed in Table 3. Assuming that histopathology accurately assesses the margin, the specificity and positive predictive value were 100% for specimen margin evaluation, whereas the sensitivity was only 14% (standard error, 4.5%) and the negative predictive value was 90% (standard error, 3.9%) (Table 2). It is interesting to note that the 1 specimen with a positive margin on both surgical pathology and cytopathology had a negative in situ margin and the patient had not developed local disease recurrence at 31 months of follow-up, but instead died of intrahepatic disease recurrence that developed away from the site of the original resection.
Table 3. Cytopathologic Results and Outcomes for Patients With a Positive Specimen Margin on Histopathology
There were no cytologically positive in situ margins identified along the cut edge of the liver remnant. Follow-up data were complete for 64 patients for median of 37.1 months (range, 12 months-56 months). Twenty patients were excluded because of perioperative deaths (6 patients; 7%) or follow-up of < 12 months. Of these 64 patients, 56 (87.5%) were found to have disease recurrence, with 34 of the patients (60.7%) developing recurrence in extrahepatic locations (lungs, brain, retroperitoneal lymph nodes, or bone) and 38 patients (67.9%) having intrahepatic recurrence, including synchronous extra- and intrahepatic recurrences. The average time to recurrence was 16.1 months. One patient (1.6%) demonstrated local recurrence at the liver resection margin in addition to multiple sites of intra- and extrahepatic recurrences. The surgical histopathology margin, cytopathologic specimen margin, and in situ cytopathologic margin were all negative for this patient (Table 3). The remaining cases of disease recurrence were either extrahepatic or intrahepatic recurrences that developed away from the site of the previous resection.
In the current study, we describe a new method for determining margin status at the time of liver resection. This method is quick (< 2 minutes) and can be universally applied using standard techniques. This approach builds on the observations by Cox et al that touch preparation cytology can be used at the time of lumpectomy for breast cancer and correlates with final histology.7 In the current study, we have shown that this approach is feasible and, given the low rate of local recurrence, may more accurately reflect the true margin status than traditional histopathology.
First, diagnostic tumor cells were identified in all cases except one in which no residual tumor was identified by histopathology. The sensitivity was still 92% for detecting viable cancer cells in 7 tumor specimens that were largely necrotic after extensive preoperative chemotherapy. Tumor sampling for cytopathology was obtained from the central portion of the bivalved tumor. Alternatively, sampling the outer edge of the tumor may have yielded viable cancer cells similar to histopathology in all these samples.
Using histopathology as the referent standard, the cytopathologic specimen margin assessment in the current study yielded a low sensitivity of 14%. This brings into question the assumption that the histopathologic evaluation of the margin is always accurate. When a liver specimen is sent to pathology, it is not always clear what constitutes the margin, particularly in the case of complex and/or nonanatomic resections. Next, fracturing of the parenchyma at the time of transection may allow for deeper penetration of ink into the cut edge of the liver. Finally, cautery artifact may distort cells at the margin. All of these can lead to under- or overestimation of margin positivity. Given the low number of local recurrences (1.6%) noted in the current study, we suggest the latter to be true. Although longer follow-up for the entire cohort of patients will allow for better assessment of the comparative accuracy of the 2 technologies, D'Angelica et al have demonstrated that 75% of local recurrences occur within 2 years of resection, which is beyond the median follow-up for the patients participating in the current study.9
The real potential benefit of using cytopathology to determine margin status is the ability to quickly assess the in situ margin without the need to remove additional tissue. The risk of local recurrence with an R1 resection has been cited by numerous authors for colorectal liver metastases and primary hepatic tumors.10–13 Because the in situ margin is arguably the most relevant margin, this technique, if validated, could prove invaluable. In the current study, no positive in situ margins were identified by cytopathology; however, 1 patient went on to develop both a local recurrence at the liver resection margin as well as intra- and extrahepatic recurrences on follow-up imaging. The multifocal nature of this patient's aggressive cancer challenges the usefulness of margin assessment by any method. Nevertheless, a potential weakness of the current study is the absence of the detection of any positive in situ margins by cytopathology, especially in cases with positive specimen margins. Although we, like others, strive for complete resection when the intention is cure, continued data acquisition will likely result in more events that will help define this methodology.
One final potential weakness of the current study is that we did not account for adjuvant chemotherapy (no patients received radiotherapy to liver resection margins), which may impact the rate of local recurrence. The current study incorporated a variety of diseases in patients who had undergone multiple pre- and postoperative therapies. Nevertheless, the vast majority of patients developed disease recurrence at some point during follow-up, thus making the argument that adjuvant therapy did not alter the risk of local recurrence.
Cytopathologic evaluation of the specimen and in situ margins after hepatectomy offers a new, quick, and feasible technique for assessing margin status with potential benefit to the patient for achieving an R0 resection. Documentation of a negative in situ margin can avoid the need for further resection or chemotherapy/radiotherapy when local recurrence is a primary concern. Continued surveillance will allow us to report the clinical impact of the in situ margin on recurrence and survival after hepatectomy.