Abstract
Objectives:
This study aimed to explore the risk factors associated with cancer cell exfoliation in Stage II and III colorectal cancer (CRC).
Methods:
This multicenter, prospective, observational study targeted 1,698 patients with cStage II and III CRC who underwent R0 resection between 2013 and 2017. Clinicopathological variables were analyzed for correlations with positive peritoneal lavage cytology (PLC).
Results:
The positive PLC rate was 2.7% (46/1,694 cases) at laparotomy and 1.6% (25/1,590 cases) after tumor resection. Logistic regression analyses identified that undifferentiated histologies diagnosed by preoperative biopsy specimen, cT4, and pN+ were independent factors that affected the positive PLC at laparotomy. The positive PLC rate at laparotomy was 4.5% (33/736 cases) among the patients with undifferentiated histology and/or cT4. Logistic regression analyses revealed that the presence of ascites and undifferentiated histology by biopsy independently affected positive PLC after tumor resection.
Conclusions:
The undifferentiated histology and/or T4 indicated by preoperative diagnosis were identified as factors affecting PLC at laparotomy. Furthermore, ascites and preoperative histological type were identified as factors affecting positive PLC after tumor resection. As factors affecting positive PLC, these preoperative findings were found to be equivalent to pathological findings.
Keywords: colorectal cancer, peritoneal lavage cytology, peritoneal metastasis, R0 resection
Introduction
Colorectal cancer (CRC) is the third most common cause of death in Japan, and it has an increasing prevalence. Complete tumor removal is the most effective treatment for CRC. However, some recurrences are inevitable after curative resections, even in cases of apparently localized tumors, and the liver, lung, and peritoneum are the most common recurrence sites[1]. Despite recent advances in knowledge regarding various clinical, biological, and pathological features associated with CRC prognosis, the depth of tumor invasion and lymph node involvement have been regarded as the most important factors affecting recurrence and prognosis. These factors are used in Dukes staging and the tumor-node-metastasis (TNM) classification system[2,3].
The prognostic values of peritoneal lavage cytology (PLC) and ascites cytology have been well established for gynecologic cancers[4,5]. Peritoneal and pleural lavage cytologies are also useful prognostic indicators for gastric, pancreatic, esophageal, and lung cancers[6-9] and contribute to patient selection for close follow-up and/or intensive chemotherapy[10-13]. On the other hand, although positive PLC has been reported to be associated with increased recurrence risk and poor prognosis[14-17], the true prognostic value of PLC for CRC has not yet been established[3,18]. In previous reports, the positive PLC rates in patients with CRC who had R0 resections were considerably low[14,15,19,20].
The identification of factors associated with cancer cell exfoliation into the peritoneal cavity must be able to clarify selection criterion for PLC. Although some previous reports examined such factors[17-21], most of them included a small number of patients, especially patients who underwent R0 resection. Furthermore, many reported factors associated with cancer cell exfoliation into the peritoneal cavity were assessed after surgery, such as pathological findings of the resected specimen.
This study aimed to explore the factors associated with cancer cell exfoliation into the peritoneal cavity using preoperative factors.
Methods
Study design
This multicenter, prospective, observational study targeted patients with CRC and was conducted by the Japanese Society for Cancer of the Colon and Rectum (JSCCR). Patients were enrolled at 21 institutions in Japan from September 2013 to December 2017. The eligibility criteria were age ≥20 years and histologically confirmed colorectal adenocarcinoma diagnosed as Stage II or III before surgery. To diagnose the CRC stage before surgery, all patients underwent barium enema, colonoscopy, abdominal ultrasound, and abdominal and chest computed tomography, regardless of the cancer site. Pelvic magnetic resonance imaging was performed for low rectal cancer. Patients with synchronous or metachronous (within 5 years) malignancies, other than carcinoma, in situ, were excluded.
Written informed consent was obtained from all patients before enrollment. The study protocol was approved by the ethics committee of the JSCCR and the local institutional review board.
Procedures
PLC was performed twice for each case: once immediately after the abdomen was opened and before the exploration and manipulation of the tumor (at laparotomy) and once after anastomosis or tumor resection (after tumor resection). PLC just after anastomosis or tumor resection was performed before peritoneal lavage. Any free fluid present in the abdominal recesses was removed for examination (ascites and peritoneal cytology). In the absence of obvious peritoneal fluid, 50 ml of physiologic saline solution was instilled into the vesicorectal/vesicouterine pouch, with the patient in a supine position. After gentle stirring, these fluids were collected. The collected lavage fluid was immediately heparinized and centrifuged at 2,000 rpm for 3 minutes, and the sediment was smeared on four glass slides and fixed with cytospray. The slides were stained using the Giemsa and Papanicolaou methods and evaluated by cytologists who were blinded to the clinical information. The smears were classified according to the Papanicolaou classification, and Class III, IV, and V were classified as positive. When both PLC and peritoneal cytology of ascites were performed, the result was classified as positive if either test was positive.
Parameters
The parameters assessed in this study were as follows: age, gender, comorbidity, neoadjuvant therapy, Eastern Cooperative Oncology Group Performance Status, location of the primary tumor, size of the primary tumor, number of tumors, presence or absence of ascites, preoperative serum levels of carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9), laparoscopic surgery, instrument anastomosis, lymph node dissection, additional resection, volume of bleeding, operative time, complications, histological type, depth of tumor invasion, lymph node metastases, lymphatic invasion, and venous invasion.
Definitions and diagnosis
Clinical and pathological data were recorded according to the JSCCR classification system[18]. The depth of tumor invasion and lymph node metastases were classified according to the eighth edition of the TNM classification system[4]. The cutoff values for CEA and CA19-9 were 5.0 ng/mL and 37.0 U/mL, respectively. For the analyses of factors affecting positive cytology, the cutoff values for body mass index (BMI) were determined using receiver operating characteristic curve analyses.
Statistical analysis
The differences between continuous and categorical variables were identified using the Mann-Whitney U test and the chi-squared test or Fisher's exact test, respectively. The factors affecting positive PLC were analyzed using binomial logistic regression analyses, and the identified factors were compared using the chi-squared test or Fisher's exact test, at the p < 0.01 level, to determine factors independently associated with positive PLC. Multivariate analyses were performed for both preoperative and pathological findings.
All data were expressed as the total number of patients and proportion of the population (%). The p-values < 0.05 were considered significant. All analyses were performed using the JMP 11 statistical software (SAS Institute Inc., Cary, NC, USA).
Results
Patient characteristics
Figure 1 shows the flow diagram used to determine patient inclusion in this study. A total of 1,785 patients with clinical Stage II and III CRC were enrolled from 21 institutions in Japan. A total of 67 patients who were not curable with surgical treatment according to the JSCCR classification system and 20 patients who had not undergone PLC at laparotomy or after tumor resection were excluded. Thus, the study population comprised 1,698 patients with CRC. PLC was performed in 1,694 patients at laparotomy and 1,590 patients after tumor resection.
Figure 1.
Flow diagram for the inclusion of patients in this study. *: R0 resection according to the JSCCR classification system **: Peritoneal lavage cytology was not performed either at laparotomy or after tumor resection.
The positive PLC rate was 2.7% (46/1,694 cases) in the patients who underwent PLC at laparotomy. The positive PLC rate was 3.0% (46/1,512 cases) in patients with pStage II and III CRC. The positive PLC rate after tumor resection was 1.6% (25/1,590 cases). Among these, the PLC at laparotomy were positive in 44% (11/25 cases). The positive PLC rate was 0.9% (14/1,544 cases) in patients with negative PLC at laparotomy and positive PLC after tumor resection.
Cutoff value of BMI
The cutoff values at laparotomy determined using receiver operating characteristic curve analyses were as follows. The area under the curve was 0.550. When the cutoff value for BMI was defined as 24, the sensitivity and specificity of BMI for predicting positive peritoneal cytology were 82.6% and 32.0%, respectively.
The cutoff values after tumor resection determined using receiver operating characteristic curve analyses were as follows. The area under the curve was 0.678. When the cutoff value for BMI was defined as 21, the sensitivity and specificity of BMI for predicting positive peritoneal cytology were 62.5% and 63.8%, respectively.
Concordance rate between pre- and postoperative findings for histology and T and N classification
Of the 1,603 patients with preoperative diagnosis of differentiated histology, 96.0% were identical with postoperative histological diagnosis, and of the 82 patients with preoperative diagnosis of undifferentiated histology, 69.5% coincided with the postoperative findings.
Of the 981 patients with cT1/T2/T3, 94.3% were pT1/T2/TT3, and of the 329 patients with cT4, 46.4% were pT4. Of the 732 patients with cN0, 69.0% were pN0, and of the 955 patients with cN+, 56.8% were pN+.
The concordance rates between the pre- and postoperative findings for histology and T and N classification were 94.7%, 74.2%, and 62.1%, respectively.
Factors affecting positive peritoneal cytology at laparotomy
Positive PLC at laparotomy was significantly associated with histology diagnosed before surgery (p = 0.003), cT (p < 0.001), pT (p < 0.001), and pN (p < 0.001), as determined by univariate analyses (Table 1).
Table 1.
Univariate Analyses of Factors Affecting Positive Peritoneal Cytology at Laparotomy for Colorectal Cancer, Which Was Curability A with Surgical Treatment According to the JSCCR Classification System.
| Variable | Category | Number | Performed
case |
Positive case
(%) |
Negative
case |
Univariate analysis
p-value |
||
|---|---|---|---|---|---|---|---|---|
| Preoperative
factor |
Gender | Male | 999 | 999 | 20 | 2.0% | 979 | 0.03 |
| Female | 698 | 694 | 26 | 3.7% | 668 | |||
| Unknown | 1 | 1 | 0 | 0.0% | 1 | |||
| Age (year old) | <70 | 864 | 860 | 26 | 3.0% | 834 | 0.43 | |
| 70≤ | 831 | 831 | 20 | 2.4% | 811 | |||
| Unknown | 3 | 3 | 0 | 0.0% | 3 | |||
| Comorbidities | + | 1086 | 1083 | 26 | 2.4% | 1057 | 0.38 | |
| - | 610 | 609 | 19 | 3.1% | 590 | |||
| Unknown | 2 | 2 | 1 | 50.0% | 1 | |||
| Neoadjuvant therapy | + | 54 | 54 | 1 | 1.9% | 53 | 0.69 | |
| - | 1644 | 1640 | 45 | 2.7% | 1595 | |||
| Performance status | 0, 1 | 1482 | 1478 | 40 | 2.7% | 1438 | 0.99 | |
| 2, 3, 4 | 148 | 148 | 4 | 2.7% | 144 | |||
| Unknown | 68 | 68 | 2 | 2.9% | 66 | |||
| Tumor location | Colon | 1268 | 1265 | 38 | 3.0% | 1228 | 0.21 | |
| Rectum | 430 | 429 | 8 | 1.9% | 421 | |||
| Histological type | Differenciated type | 1529 | 1526 | 35 | 2.3% | 1491 | 0.003 | |
| Undifferenciated type | 77 | 76 | 6 | 7.9% | 70 | |||
| Unknown | 92 | 92 | 5 | 5.4% | 87 | |||
| Depth of invasion | T1, T2, T3 | 988 | 984 | 13 | 1.3% | 971 | <0.001 | |
| T4a, T4b | 710 | 710 | 33 | 4.6% | 677 | |||
| Unknown | 0 | 0 | 0 | 0.0% | 0 | |||
| LN metastases | - | 736 | 734 | 12 | 1.6% | 722 | 0.02 | |
| + | 959 | 957 | 34 | 3.6% | 923 | |||
| Unknown | 3 | 3 | 0 | 0.0% | 3 | |||
| Number of tumor | Solitary | 1614 | 1610 | 42 | 2.6% | 1568 | 0.19 | |
| Multiple | 79 | 79 | 4 | 5.1% | 75 | |||
| Unknown | 5 | 5 | 0 | 0.0% | 5 | |||
| Size of tumor (mm) | <60 | 1183 | 1179 | 24 | 2.0% | 963 | 0.04 | |
| 60≤ | 505 | 505 | 22 | 4.4% | 483 | |||
| Unknown | 10 | 10 | 0 | 0.0% | 10 | |||
| Ascites | + | 94 | 94 | 4 | 4.3% | 90 | 0.34 | |
| - | 1604 | 1600 | 42 | 2.6% | 1558 | |||
| Serum CEA level | Low | 1055 | 1051 | 21 | 2.0% | 1030 | 0.04 | |
| High | 597 | 597 | 22 | 3.7% | 575 | |||
| Unknown | 46 | 46 | 3 | 6.5% | 43 | |||
| Serum CA19-9 level | Low | 1456 | 1452 | 35 | 2.4% | 1417 | 0.12 | |
| High | 185 | 185 | 8 | 4.3% | 177 | |||
| Unknown | 57 | 57 | 3 | 5.3% | 54 | |||
| Operative
factor |
Laparoscopic surgey | + | 667 | 663 | 13 | 2.0% | 650 | 0.13 |
| - | 1031 | 1031 | 33 | 3.2% | 998 | |||
| Histological
factor |
Histological type | Differenciated type | 1567 | 1564 | 40 | 2.6% | 1524 | 0.12 |
| Undifferenciated type | 122 | 121 | 6 | 5.0% | 115 | |||
| Unknown | 9 | 9 | 0 | 0.0% | 9 | |||
| Depth of invasion | T1, T2, T3 | 1305 | 1301 | 21 | 1.6% | 1280 | <0.001 | |
| T4a, T4b | 385 | 385 | 25 | 6.5% | 360 | |||
| Unknown | 8 | 8 | 0 | 0.0% | 8 | |||
| LN metastasis | + | 772 | 771 | 33 | 4.3% | 738 | <0.001 | |
| - | 922 | 919 | 13 | 1.4% | 906 | |||
| Unknown | 4 | 4 | 0 | 0.0% | 4 | |||
| Lymphatic invasion | 0 | 619 | 618 | 10 | 1.6% | 608 | 0.04 | |
| 1, 2, 3 | 1070 | 1067 | 35 | 3.3% | 1032 | |||
| Unknown | 9 | 9 | 1 | 11.1% | 8 | |||
| Venous invasion | 0 | 494 | 493 | 9 | 1.8% | 484 | 0.16 | |
| 1, 2, 3 | 1192 | 1189 | 36 | 3.0% | 1153 | |||
| Unknown | 12 | 12 | 1 | 8.3% | 11 | |||
JSCCR: Japanese Society for Cancer of the Colon and Rectum LN: lymph node
When using the cT, logistic regression analyses revealed that preoperative histology, cT, and pN independently affected positive PLC at laparotomy (Table 2). When using the pT, logistic regression analyses revealed that preoperative histology (hazard ratio: 2.70, p-value < 0.04), pT (hazard ratio: 3.20, p-value < 0.001), and pN (hazard ratio: 2.70, p-value = 0.01) independently affected positive PLC at laparotomy.
Table 2.
Multivariate Regression Analysis for Factors Affecting Positive Peritoneal Cytology at Laparotomy for Colorectal Cancer, Which Was Curability A with Surgical Treatment According to the JSCCR Classification System.
| Variables selected | Hazard ratio | 95% confidence interval | p-value |
|---|---|---|---|
| Histological type (before surgery) | 3.02 | 1.20–7.58 | 0.020 |
| Depth of invasion (before surgery) | 2.67 | 1.36–5.24 | 0.004 |
| Lymph node metastasis (histological) | 3.10 | 1.54–6.27 | 0.002 |
JSCCR: Japanese Society for Cancer of the Colon and Rectum
Factors affecting positive peritoneal cytology after tumor resection
Positive PLC after tumor resection was significantly associated with preoperative histology (p = 0.005), ascites (p = 0.004), pT (p = 0.004), and pN (p = 0.004) according to the univariate analyses (Table 3). Logistic regression analyses revealed that, among these factors, ascites and preoperative histology independently affected positive PLC after tumor resection (Table 3).
Table 3.
Univariate and Multivariate Analyses of Factors Affecting Positive Peritoneal Lavage Cytology after Tumor Resection for All Colorectal Cancer, Which Was Curability A with Surgical Treatment According to the JSCCR Classification System.
| Variable | Category | Number | Performed
case |
Positive case | Negative
case |
Univariate
analysis |
Multivariate analysis | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Number | % | Number | p-value | HR | 95% CI | p-value | |||||
| Preoperative
factor |
Gender | Male | 999 | 937 | 15 | 1.6% | 922 | 0.91 | |||
| Female | 698 | 652 | 10 | 1.5% | 642 | ||||||
| Unknown | 1 | 1 | 0 | 0.0% | 1 | ||||||
| Age (year old) | <70 | 864 | 819 | 13 | 1.6% | 806 | 0.97 | ||||
| 70≤ | 831 | 768 | 12 | 1.6% | 756 | ||||||
| Unknown | 3 | 3 | 0 | 0.0% | 3 | ||||||
| Comorbidities | + | 1086 | 1017 | 14 | 1.4% | 1003 | 0.56 | ||||
| - | 640 | 571 | 10 | 1.8% | 561 | ||||||
| Unknown | 2 | 2 | 1 | 50.0% | 1 | ||||||
| Neoadjuvant therapy | + | 54 | 51 | 0 | 0.0% | 51 | 0.36 | ||||
| - | 1644 | 1539 | 25 | 1.6% | 1514 | ||||||
| Performance status | 0, 1 | 1482 | 1384 | 20 | 1.4% | 1364 | 0.055 | ||||
| 2, 3, 4 | 148 | 138 | 5 | 3.6% | 133 | ||||||
| Unknown | 68 | 68 | 0 | 0.0% | 68 | ||||||
| Tumor location | Colon | 1268 | 1176 | 21 | 1.8% | 1155 | 0.240 | ||||
| Rectum | 430 | 414 | 4 | 1.0% | 410 | ||||||
| Histological type | Differenciated type | 1529 | 1441 | 20 | 1.4% | 1421 | 0.005 | 3.39 | 1.08–10.61 | 0.04 | |
| Undifferenciated type | 77 | 70 | 4 | 5.7% | 66 | ||||||
| Unknown | 92 | 79 | 1 | 1.3% | 78 | ||||||
| Depth of invasion | T1, T2, T3 | 988 | 957 | 13 | 1.4% | 944 | 0.40 | ||||
| T4a, T4b | 710 | 633 | 12 | 1.9% | 621 | ||||||
| Unknown | 0 | 0 | 0 | 0.0% | 0 | ||||||
| LN metastases | - | 736 | 704 | 9 | 1.3% | 695 | |||||
| + | 959 | 957 | 16 | 1.7% | 941 | 0.40 | |||||
| Unknown | 3 | 3 | 0 | 0.0% | 3 | ||||||
| Number of tumor | Solitary | 1613 | 1512 | 23 | 1.5% | 1489 | 0.40 | ||||
| Multiple | 79 | 72 | 2 | 2.8% | 70 | ||||||
| Unknown | 5 | 5 | 0 | 0.0% | 5 | ||||||
| Size of tumor (mm) | <60 | 1183 | 1124 | 13 | 1.2% | 1111 | 0.03 | ||||
| 60≤ | 505 | 457 | 12 | 2.6% | 445 | ||||||
| Unknown | 10 | 9 | 0 | 0.0% | 9 | ||||||
| Ascites | + | 94 | 68 | 4 | 5.9% | 64 | 0.004 | 3.86 | 1.25–11.95 | 0.02 | |
| - | 1604 | 1522 | 21 | 1.4% | 1501 | ||||||
| Serum CEA level | Low | 1055 | 1010 | 15 | 1.5% | 995 | 0.99 | ||||
| High | 597 | 539 | 8 | 1.5% | 531 | ||||||
| Unknown | 46 | 41 | 2 | 4.9% | 39 | ||||||
| Serum CA19-9 level | Low | 1456 | 1366 | 21 | 1.5% | 1345 | 0.70 | ||||
| High | 185 | 173 | 2 | 1.2% | 171 | ||||||
| Unknown | 57 | 51 | 2 | 3.9% | 49 | ||||||
| Operative
factor |
Laparoscopic surgey | + | 667 | 647 | 12 | 1.9% | 635 | 0.45 | |||
| - | 1031 | 943 | 13 | 1.4% | 930 | ||||||
| Instrument anastomosis | + | 1149 | 1062 | 18 | 1.7% | 1044 | 0.60 | ||||
| - | 539 | 519 | 7 | 1.3% | 512 | ||||||
| Unknown | 10 | 9 | 0 | 0.0% | 9 | ||||||
| LN dissection | D0, D1 | 17 | 11 | 1 | 9.1% | 10 | 0.06 | ||||
| D2, D3 | 1676 | 1574 | 24 | 1.5% | 1550 | ||||||
| Unknown | 5 | 5 | 0 | 0.0% | 5 | ||||||
| Additional resection | + | 159 | 149 | 1 | 0.7% | 148 | 0.35 | ||||
| - | 1528 | 1431 | 24 | 1.7% | 1407 | ||||||
| Unknown | 11 | 10 | 0 | 0.0% | 10 | ||||||
| Bleeding volume (ml) | 175≤ | 474 | 428 | 3 | 0.7% | 425 | 0.09 | ||||
| <175 | 1219 | 1157 | 22 | 1.9% | 1135 | ||||||
| Unknown | 5 | 5 | 0 | 0.0% | 5 | ||||||
| Operative time (min) | 226≤ | 856 | 814 | 15 | 1.8% | 799 | 0.38 | ||||
| <226 | 840 | 774 | 10 | 1.3% | 764 | ||||||
| Unknown | 2 | 2 | 0 | 0.0% | 2 | ||||||
| Complication (all) | + | 459 | 428 | 9 | 2.1% | 419 | 0.30 | ||||
| - | 1239 | 1162 | 16 | 1.4% | 1146 | ||||||
| Complication (Grade 3 -) | + | 184 | 169 | 3 | 1.8% | 166 | 0.83 | ||||
| - | 1514 | 1421 | 22 | 1.5% | 1399 | ||||||
| Histological
factor |
Histological type | Differenciated type | 1567 | 1481 | 21 | 1.4% | 1460 | 0.045 | |||
| Undifferenciated type | 122 | 100 | 4 | 4.0% | 96 | ||||||
| Unknown | 9 | 9 | 0 | 0.0% | 9 | ||||||
| Depth of invasion | T1, T2, T3 | 1305 | 1255 | 14 | 1.1% | 1241 | 0.004 | 2.13 | 0.89–5.06 | 0.09 | |
| T4a, T4b | 385 | 327 | 11 | 3.4% | 316 | ||||||
| Unknown | 8 | 8 | 0 | 0.0% | 8 | ||||||
| LN metastasis | + | 772 | 713 | 14 | 2.0% | 699 | 0.004 | 1.41 | 0.61–3.28 | 0.42 | |
| - | 922 | 873 | 11 | 1.3% | 862 | ||||||
| Unknown | 4 | 4 | 0 | 0.0% | 4 | ||||||
| Lymphatic invasion | 0 | 619 | 601 | 6 | 1.0% | 595 | 0.15 | ||||
| 1, 2, 3 | 1070 | 980 | 19 | 1.9% | 961 | ||||||
| Unknown | 9 | 9 | 0 | 0.0% | 9 | ||||||
| Venous invasion | 0 | 494 | 475 | 5 | 1.1% | 470 | 0.27 | ||||
| 1, 2, 3 | 1192 | 1103 | 20 | 1.8% | 1083 | ||||||
| Unknown | 12 | 12 | 0 | 0.0% | 12 | ||||||
JSCCR: Japanese Society for Cancer of the Colon and Rectum LN: lymph node
Among the patients with negative PLC at laparotomy, positive PLC after tumor resection was significantly associated with ascites (p < 0.001) and preoperative histology (p = 0.002) according to the univariate analyses (Table 4). In the logistic regression analyses, among these two factors, both were identified as independent factors affecting positive PLC after tumor resection (Table 4).
Table 4.
Univariate and Multivariate Analyses of Factors Affecting Positive Peritoneal Lavage Cytology after Tumor Resection for Colorectal Cancer, Which Was Curability A with Surgical Treatment According to the JSCCR Classification System among the Cases with Negative Peritoneal Cytology at Laparotomy.
| Variable | Category | Number | Performed
case |
Positive case | Negative
case |
Univariate
analysis |
Multivariate analysis | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Number | % | Number | p-value | HR | 95% CI | p-value | |||||
| Preoperative
factor |
Gender | Male | 999 | 918 | 10 | 1.1% | 908 | 0.35 | |||
| Female | 698 | 625 | 4 | 0.6% | 621 | ||||||
| Unknown | 1 | 1 | 0 | 0.0% | 1 | ||||||
| Age (year old) | <70 | 864 | 790 | 6 | 0.8% | 784 | 0.52 | ||||
| 70≤ | 831 | 751 | 8 | 1.1% | 743 | ||||||
| Unknown | 3 | 3 | 0 | 0.0% | 3 | ||||||
| Comorbidities | + | 1086 | 990 | 8 | 0.8% | 982 | 0.58 | ||||
| - | 640 | 553 | 6 | 1.1% | 547 | ||||||
| Unknown | 2 | 1 | 0 | 0.0% | 1 | ||||||
| Neoadjuvant therapy | + | 54 | 50 | 0 | 0.0% | 50 | 0.49 | ||||
| - | 1644 | 1494 | 14 | 0.9% | 1480 | ||||||
| Performance status | 0, 1 | 1482 | 1344 | 10 | 0.7% | 1334 | 0.02 | ||||
| 2, 3, 4 | 148 | 134 | 4 | 3.0% | 130 | ||||||
| Unknown | 68 | 66 | 0 | 0.0% | 66 | ||||||
| Tumor location | Colon | 1268 | 1137 | 11 | 1.0% | 1126 | 0.68 | ||||
| Rectum | 430 | 407 | 3 | 0.7% | 404 | ||||||
| Histological type | Differenciated type | 1529 | 1406 | 11 | 0.8% | 1395 | 0.002 | 5.83 | 1.53–22.2 | 0.001 | |
| Undifferenciated type | 77 | 63 | 3 | 4.8% | 60 | ||||||
| Unknown | 92 | 75 | 0 | 0.0% | 75 | ||||||
| Depth of invasion | T1, T2, T3 | 988 | 943 | 11 | 1.2% | 932 | 0.18 | ||||
| T4a, T4b | 710 | 601 | 3 | 0.5% | 598 | ||||||
| Unknown | 0 | 0 | 0 | 0.0% | 0 | ||||||
| LN metastases | - | 736 | 692 | 5 | 0.7% | 687 | 0.42 | ||||
| + | 959 | 798 | 9 | 1.1% | 789 | ||||||
| Unknown | 3 | 3 | 0 | 0.0% | 3 | ||||||
| Number of tumor | Solitary | 1613 | 1470 | 14 | 1.0% | 1456 | 0.42 | ||||
| Multiple | 79 | 68 | 0 | 0.0% | 68 | ||||||
| Unknown | 5 | 5 | 0 | 0.0% | 5 | ||||||
| Size of tumor (mm) | <60 | 1183 | 1097 | 9 | 0.8% | 1088 | 0.55 | ||||
| 60≤ | 505 | 438 | 5 | 1.1% | 433 | ||||||
| Unknown | 10 | 9 | 0 | 0.0% | 9 | ||||||
| Ascites | + | 94 | 66 | 4 | 6.1% | 62 | <0.0001 | 8.62 | 2.58–28.8 | 0.0005 | |
| - | 1604 | 1478 | 10 | 0.7% | 1468 | ||||||
| Serum CEA level | Low | 1055 | 987 | 8 | 0.8% | 979 | 0.75 | ||||
| High | 597 | 519 | 5 | 1.0% | 514 | ||||||
| Unknown | 46 | 38 | 1 | 2.6% | 37 | ||||||
| Serum CA19-9 level | Low | 1456 | 1331 | 11 | 0.8% | 1320 | 0.61 | ||||
| High | 185 | 165 | 2 | 1.2% | 163 | ||||||
| Unknown | 57 | 48 | 1 | 2.1% | 47 | ||||||
| Operative
factor |
Laparoscopic surgey | + | 667 | 632 | 9 | 1.4% | 623 | 0.07 | |||
| - | 1031 | 912 | 5 | 0.5% | 907 | ||||||
| Instrument anastomosis | + | 1149 | 1030 | 11 | 1.1% | 1019 | 0.35 | ||||
| - | 539 | 506 | 3 | 0.6% | 503 | ||||||
| Unknown | 10 | 8 | 0 | 0.0% | 8 | ||||||
| LN dissection | D0, D1 | 17 | 10 | 0 | 0.0% | 10 | 0.75 | ||||
| D2, D3 | 1676 | 1529 | 14 | 0.9% | 1515 | ||||||
| Unknown | 5 | 5 | 0 | 0.0% | 5 | ||||||
| Additional resection | + | 159 | 146 | 1 | 0.7% | 145 | 0.76 | ||||
| - | 1528 | 1388 | 13 | 0.9% | 1375 | ||||||
| Unknown | 11 | 10 | 0 | 0.0% | 10 | ||||||
| Bleeding volume (ml) | 175≤ | 474 | 420 | 3 | 0.7% | 417 | 0.62 | ||||
| <175 | 1219 | 1119 | 11 | 1.0% | 1108 | ||||||
| Unknown | 5 | 5 | 0 | 0.0% | 5 | ||||||
| Operative time (min) | 226≤ | 856 | 791 | 9 | 1.1% | 782 | 0.33 | ||||
| <226 | 840 | 751 | 5 | 0.7% | 746 | ||||||
| Unknown | 2 | 2 | 0 | 0.0% | 2 | ||||||
| Complication (all) | + | 459 | 412 | 5 | 1.2% | 407 | 0.44 | ||||
| - | 1239 | 1132 | 9 | 0.8% | 1123 | ||||||
| Complication (Grade 3 -) | + | 184 | 159 | 1 | 0.6% | 158 | 0.69 | ||||
| - | 1514 | 1385 | 13 | 0.9% | 1372 | ||||||
| Histological
factor |
Histological type | Differenciated type | 1567 | 1442 | 12 | 0.8% | 1430 | 0.20 | |||
| Undifferenciated type | 122 | 93 | 2 | 2.2% | 91 | ||||||
| Unknown | 9 | 9 | 0 | 0.0% | 9 | ||||||
| Depth of invasion | T1, T2, T3 | 1305 | 1231 | 9 | 0.7% | 1222 | 0.14 | ||||
| T4a, T4b | 385 | 305 | 5 | 1.6% | 300 | ||||||
| Unknown | 8 | 8 | 0 | 0.0% | 8 | ||||||
| LN metastasis | + | 772 | 680 | 5 | 0.7% | 675 | 0.53 | ||||
| - | 922 | 860 | 9 | 1.0% | 851 | ||||||
| Unknown | 4 | 4 | 0 | 0.0% | 4 | ||||||
| Lymphatic invasion | 0 | 619 | 590 | 5 | 0.8% | 585 | 0.84 | ||||
| 1, 2, 3 | 1070 | 946 | 9 | 1.0% | 937 | ||||||
| Unknown | 9 | 8 | 0 | 0.0% | 8 | ||||||
| Venous invasion | 0 | 494 | 465 | 1 | 0.2% | 464 | 0.06 | ||||
| 1, 2, 3 | 1192 | 1068 | 13 | 1.2% | 1055 | ||||||
| Unknown | 12 | 11 | 0 | 0.0% | 11 | ||||||
JSCCR: Japanese Society for Cancer of the Colon and Rectum LN: lymph node
Positive PLC rates according to factors affecting positive peritoneal cytology
Table 5 shows the positive PLC rates, according to the factors affecting positive peritoneal cytology results at laparotomy. At least one of the identified factors that could be assessed before surgery (preoperative histology, cT4) were identified in 736 patients (43.4%) of the 1,694 patients who had PLC at laparotomy, and the positive PLC rate was 4.5% (33/1,694 cases). Furthermore, at least one of the identified factors including pN+ was identified in 1,131 patients (66.8%), and the positive PLC was 3.8% (43 cases). At least one of the identified factors (preoperative histology, pT4, and pN) was identified in 949 patients (56.0%), and the positive PLC rate was 4.4% (42/949 cases). Among the 13 patients who had positive PLC at laparotomy, despite having no factors affecting positive PLC at laparotomy, which were possible to know before surgery, four patients were Stage IIIa, six were Stage IIIb, and three were Stage group II based on both pre- and postoperative information.
Table 5.
Positive Rates of Peritoneal Cytology According to Factors Affecting Positive Peritoneal Cytology at Laparotomy.
| Performed case | Positive case (%) | ||||
|---|---|---|---|---|---|
| #1 | Histological type (preoperative) | Undif type | 76 | 6 | (7.9) |
| #2 | Depth of invasion (preoperative) | T4 | 710 | 33 | (4.6) |
| #3 | Depth of invasion (histological) | T4 | 385 | 25 | (6.5) |
| #4 | Lymph node metastasis (histological) | + | 771 | 33 | (4.3) |
| #1 or #2 | 736 | 33 | (4.5) | ||
| #1 or #2 or #4 | 1131 | 43 | (3.8) | ||
| #1 or #3 or #4 | 949 | 42 | (4.4) | ||
| All cases | 1,694 | 46 | (2.7) | ||
Undif type: Undiferenciated type
Among the 1,590 patients who had PLC after tumor resection, ascites and/or preoperative undifferentiated histology, which were factors identified to affect positive PLC after tumor resection, were identified in 8.4% (133/1,590 cases), and the positive PLC rate among these was 5.3% (7/133 cases). Among the 1,544 patients who had negative PLC at laparotomy and had PLC after tumor resection, ascites and/or preoperative undifferentiated histology were identified in 8.0% (124/1,544 cases), and the positive PLC rate among these was 4.8% (6/124 cases).
Discussion
The major spreading routes of CRC are hematogenous, lymphogenous, and peritoneal dissemination[22]. Although peritoneal metastasis is one of the most common sites of recurrence following liver and lung metastasis, it is less frequent and considered to be less prognostically important than other routes[23]. Diagnosing peritoneal metastasis can be difficult because of the multiplicity and microscopic sizes of the implants, although peritoneal metastasis can result in intractable ascites, intestinal obstructions, and further tumor proliferation, and peritoneal metastasis must often be diagnosed by surgery[24]. Peritoneal recurrence is often diagnosed at terminal stages, and most patients with peritoneal metastasis have a poor prognosis[24]. Therefore, identifying the well-defined risk factors that can predict peritoneal recurrence during early stages may facilitate the treatment of high-risk patients using strategies such as more intensive postoperative surveillance and adjuvant chemotherapy that may improve prognoses.
The presence of free cancer cells in the abdominal cavity is thought to represent a precursor state for peritoneal dissemination. In CRC, positive lavage cytology has been associated with poor oncological outcomes[14-17], although its significance on the prognosis has not been fully established[3,18]. The reported positive rate of PLC ranged from 2.2% to 20.0%[14,15,17,19,20,25], and this large range may be due to the differences in patient populations, lavage methods, or criteria used for assessment. In many of these reports, the numbers of enrolled cases were small and likely insufficient to reach definitive conclusions. Furthermore, T2 cancer, which rarely progresses to positive peritoneal cytology, and Stage IV CRC, including peritoneal metastasis cases that are highly likely to progress to positive peritoneal cytology, were included in many of these studies. In addition, most reports included PLC at laparotomy; however, none of these reports included PLC after tumor resection, although cancer cells might exfoliate into the peritoneal cavity during the surgical procedure[26,27]. The present study assessed the incidence of free cancer cells and the factors affecting the exfoliation of cancer cells into the peritoneal cavity, by performing peritoneal cytology before and after curative resections in patients with CRC diagnosed as Stage II and III.
In our series, peritoneal cytology at laparotomy was positive in 2.7% of all patients and 3.0% of patients with pathological Stage II and III. The positive peritoneal cytology rates at laparotomy have been reported as 3.1% in T3 and T4 CRC with curative surgery by Nishikawa et al.[14], as 3.6% in Stage II and III CRC by Noura et al.[19], and as 4.6% in T3 and T4 CRC by Fujii et al.[20]. The positive PLC rate in Stage II and III CRC, for which the clinical utility of peritoneal cytology at laparotomy would be expected, was assumed to be approximately 3%-4.6%. The previously reported factors that were significantly correlated with positive PLC at laparotomy are as follows: peritoneal dissemination, liver metastasis, ascites, curability, histological type, depth of invasion (pT), lymph node metastasis (pN), lymphatic invasion, and venous invasion[14,17,19-21]. However, the factors that have been identified, including pathological findings, represent information that can only be obtained after surgery and cannot be used to determine whether PLC should be performed at laparotomy. Furthermore, PLC may have little clinical significance for Stage IV CRC because the treatment strategies, including postoperative chemotherapy and follow-up, would not be altered by the PLC results. In the present study, statistical analyses were performed for both preoperative and pathological findings for identifying preoperative factors affecting positive PLC equivalent to pathological findings. Additionally, preoperative histological type and cT were identified as factors affecting positive PLC at laparotomy and pT and pN for Stage II and III CRC with R0 resection. Preoperative histological type by the biopsy tissue was thought as an affecting factor of positive PLC at laparotomy because histological type as final diagnosis was classified by the most predominant histological type in a lesion, not by lower differentiated histological type. Furthermore, it was thought that cT and cN were not affecting factors of positive PLC at laparotomy because the accuracies of their preoperative diagnosis were low.
The economic and clinical utilities of PLC would increase if this procedure could be performed only in those cases for which the treatment strategy depends on the PLC and that presents with factors affecting positive PLC. In the present study, preoperative histological type and cT that could be identified before surgery were identified as affecting factors for positive PLC at laparotomy. When peritoneal cytology was performed on cases that presented with at least one factor among these two identified preoperative factors, PLC could be omitted in almost 55% of all patients, and the positive rate increased from 2.7% to 4.5%. Of the 46 patients with positive PLC at laparotomy, two high-risk factors that were possible to know before surgery did not effectively identify 28% of patients. Among them, three patients with Stage II CRC, for whom the treatment strategy such as the use of adjuvant chemotherapy may be affected by the PLC, were included, and these three patients did not have any factors affecting positive PLC, including pathological factors. Based on these results, these preoperative factors may be useful for selecting promising candidates to undergo PLC at laparotomy, for whom PLC could provide useful information to determine proper treatment strategies.
In the present study, PLC after tumor resection was positive in 1.6% of patients with Stage II and III and in 0.9% of patients with Stage II and III in negative PLC at laparotomy. More than half of the cases with positive PLC after tumor resection presented negative peritoneal cytology at laparotomy, suggesting that cancer cells were exfoliated into the peritoneal cavity during surgical manipulations in some cases[26,27]. Free malignant cells may also spill out from the intestinal lumen[27] or damaged lymphatics[26]. The factors that were identified to affect positive PLC after tumor resection included preoperative parameters, such as ascites and preoperative histology, even when the postoperative information, such as pathological findings, were included in the analysis regardless of PLC at laparotomy. In the present study, PLC after tumor resection could be omitted in more than 90% of all patients by selecting patients according to the factor affecting positive PLC after tumor resection, while increasing a positive rate from 1.6% to 5.3%. The utility of PLC according to the high-risk factors was thought to be in doubt because there were few positive cases in the selected cases according to the high-risk factors in spite of the rise in positive rate. However, this result is based on the outcome of peritoneal cytology before performing peritoneal lavage after tumor resection and might be improved by performing sufficient peritoneal lavage to remove exfoliated cancer cells. Therefore, the factors affecting positive PLC after tumor resection identified in the present study may not be associated with oncological results, such as recurrence and prognosis. No oncological results associated with PLC after tumor resection have yet been reported. However, sufficient peritoneal lavage following tumor resection may be necessary for high-risk patients to remove cancer cells that have exfoliated to the peritoneal cavity. Further study of PLC after tumor resections remains necessary to validate the clinical significance, sampling method, and factors affecting positive PLC. The development of strategies designed to eliminate the exfoliation of cancer cells into the peritoneal cavity is also necessary.
This study has some limitations. First, the sample size, especially the positive cytology cases, was relatively small. Second, it is necessary for the clinical significance of PLC to confirm oncological outcomes, although the present study suggested that we can efficiently obtain clinically useful information regarding the treatment strategy by selectively performing peritoneal cytology at laparotomy on patients that present factors that were possible to know before surgery. Third, the effect of intraperitoneal lavage for eliminating exfoliated cancer cells could not be properly evaluated in the current study since performing PLC just after peritoneal lavage could not be regulated according to the protocol.
In conclusion, we found that the undifferentiated histology and/or T4 indicated by preoperative diagnosis were identified as factors affecting PLC at laparotomy. Furthermore, ascites and preoperative histological type were identified as factors affecting positive PLC after tumor resection. As factors affecting positive PLC, these preoperative findings were found to be equivalent to pathological findings.
Conflicts of Interest
There are no conflicts of interest.
Author Contributions
Harunobu Sato, Kenjiro Kotake, Kotaro Maeda, Hirotoshi Kobayashi, Hiroshi Takahashi, and Kenichi Sugihara meet all the criteria for authorship as per the ICMJE recommendations and substantially contributed to the manuscript. Harunobu Sato is the corresponding author.
Approval by Institutional Review Board (IRB)
This study was approved by the Ethics Committee of the Fujita Health University (Number 13-125).
Acknowledgements
This paper was presented on behalf of the Study Group for Peritoneal Metastasis from Colorectal Cancer by the Japanese Society for Cancer of the Colon and Rectum. The authors thank the following institutional investigators for collecting patient data: Yusuke Kinugasa (Tokyo Medical and Dental University), Ichiro Takemasa (Sapporo Medical University), Hajime Morohashi (Hirosaki University), Toshifumi Wakai (Niigata University), Kazuo Hase (National Defense Medical University), Soichiro Ishihara (Tokyo University), Yukihide Kanemitsu (National Cancer Center Hospital), Michio Itabashi (Tokyo Women's Medical University), Yojiro Hashiguchi (Teikyo University), Koji Komori (Aichi Cancer Center Hospital), Shinsuke Kazama (Saitama Cancer Center), Kimihiko Funahashi (Toho University, Omori Medical Center), Hideyuki Ishida (Saitama Medical Center), Tsuyoshi Sutoh (Yamagata Prefectural Central Hospital), Keiji Koda (Teikyo University Medical Center), Masao Kameyama (Bell Land General Hospital), and Masayuki Ohue (Osaka International Cancer Institute).
References
- 1.Hashiguchi Y, Muro K, Saito Y, et al. Japanese Society for Cancer of the Colon and Rectum (JSCCR) guidelines 2019 for the treatment of colorectal cancer. Int J Clin Oncol. 2020 Jan; 25(1): 1-42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Dukes CE. The classification of cancer of the rectum. J Pathol Bacteriol. 1932 Jun; 35(3): 323-32. [Google Scholar]
- 3.Brierley JD, Gospodarowicz MK, Wittekind C. TNM cassification of malignant tumors. UICC (Union for International Cancer Control). 8th ed. John Wiley & Sons. Ltd; 2017. 73-6 p. [Google Scholar]
- 4.Ziselman EM, Harkavy SE, Hogan M, et al. Peritoneal washing cytology. Uses and diagnostic criteria in gynecologic neoplasms. Acta Cytol. 1984 Mar-Apr; 28(2): 105-10. [PubMed] [Google Scholar]
- 5.Lowe E, McKenna H. Peritoneal washing cytology: a retrospective analysis of 175 gynaecological patients. Aust NZ J Obstet Gynaecol. 1989 Feb; 29(1): 55-61. [DOI] [PubMed] [Google Scholar]
- 6.Miyashiro I, Takachi K, Doki Y, et al. When is curative gastrectomy justified for gastric cancer with positive peritoneal lavage cytology but negative macroscopic peritoneal implant? World J Surg. 2005 Sep; 29(9): 1131-4. [DOI] [PubMed] [Google Scholar]
- 7.Ishikawa O, Wada H, Ohigashi H, et al. Postoperative cytology for drained fluid from the pancreatic bed after “curative” resection of pancreatic cancers: does it predict both the patient's prognosis and the site of cancer recurrence? Ann Surg. 2003 Jul; 238(1): 103-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Doki Y, Kabuto T, Ishikawa O, et al. Does pleural lavage cytology before thoracic closure predict both patient's prognosis and site of cancer recurrence after resection of esophageal cancer? Surgery. 2001 Nov; 130(5): 792-7. [DOI] [PubMed] [Google Scholar]
- 9.Higashiyama M, Doi O, Kodama K, et al. Pleural lavage cytology immediately after thoracotomy and before closure of the thoracic cavity for lung cancer without pleural effusion and dissemination: clinicopathologic and prognostic analysis. Ann Surg Oncol. 1997 Jul-Aug; 4(4): 409-15. [DOI] [PubMed] [Google Scholar]
- 10.Martin JK Jr, Goellner JR. Abdominal fluid cytology in patients with gastrointestinal malignant lesions. Mayo Clin Proc. 1986 Jun; 61(6): 467-71. [DOI] [PubMed] [Google Scholar]
- 11.Vogel P, Rüschoff J, Kümmel S, et al. Prognostic value of microscopic peritoneal dissemination: comparison between colon and gastric cancer. Dis Colon Rectum. 2000 Jan; 43(1): 92-100. [DOI] [PubMed] [Google Scholar]
- 12.Bonenkamp JJ, Songun I, Hermans J, et al. Prognostic value of positive cytology findings from abdominal washings in patients with gastric cancer. Br J Surg. 1996 May; 83(5): 672-4. [DOI] [PubMed] [Google Scholar]
- 13.Nakajima T, Harashima S, Hirata M, et al. Prognostic and therapeutic values of peritoneal cytology in gastric cancer. Acta Cytol. 1978 Jul-Aug; 22(4): 225-9. [PubMed] [Google Scholar]
- 14.Nishikawa T, Watanabe T, Sunami E, et al. Prognostic value of peritoneal cytology and the combination of peritoneal cytology and peritoneal dissemination in colorectal cancer. Dis Colon Rectum. 2009 Dec; 52(12): 2016-21. [DOI] [PubMed] [Google Scholar]
- 15.Yamamoto S, Akasu T, Fujita S, et al. Long-term prognostic value of conventional peritoneal cytology after curative resection for colorectal carcinoma. Jpn J Clin Oncol. 2003 Jan; 33(1): 33-7. [DOI] [PubMed] [Google Scholar]
- 16.Rekhraj S, Aziz O, Prabhudesai S, et al. Can intra-operative intraperitoneal free cancer cell detection techniques identify patients at higher recurrence risk following curative colorectal cancer resection: a meta-analysis. Ann Surg Oncol. 2008 Jan; 15(1): 60-8. [DOI] [PubMed] [Google Scholar]
- 17.Katoh H, Yamashita K, Sato T, et al. Prognostic significance of peritoneal tumour cells identified at surgery for colorectal cancer. Br J Surg. 2009 Jul; 96(7): 769-77. [DOI] [PubMed] [Google Scholar]
- 18.Japan Society for Cancer of the Colon and Rectum. Japanese classification of colorectal carcinoma: the 3d English edition [Secondary publication]. J Anus Rectum Colon. 2019 Oct; 3(4): 175-95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Noura S, Ohue M, Seki Y, et al. Long-term prognostic value of conventional peritoneal lavage cytology in patients undergoing curative colorectal cancer resection. Dis Colon Rectum. 2009 Jul; 52(7): 1312-20. [DOI] [PubMed] [Google Scholar]
- 20.Fujii S, Shimada H, Yamagishi S, et al. Evaluation of intraperitoneal lavage cytology before colorectal cancer resection. Int J Colorectal Dis. 2009 Aug; 24(8): 907-14. [DOI] [PubMed] [Google Scholar]
- 21.Gozalan U, Yasti AC, Yuksek YN, et al. Peritoneal cytology in colorectal cancer: incidence and prognostic value. Am J Surg. 2007 Jun; 193(6): 672-5. [DOI] [PubMed] [Google Scholar]
- 22.Welch JP, Donaldson GA. The clinical correlation of an autopsy study of recurrent colorectal cancer. Ann Surg. 1979 Apr; 189(4): 496-502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Sadahiro S, Suzuki T, Ishikawa K, et al. Recurrence patterns after curative resection of colorectal cancer in patients followed for a minimum of ten years. Hepatogastroenterology. 2003 Sep-Oct; 50(53): 1362-6. [PubMed] [Google Scholar]
- 24.Yamaguchi A, Tsukioka Y, Fushida S, et al. Intraperitoneal hyperthermic treatment for peritoneal dissemination of colorectal cancers. Dis Colon Rectum. 1992 Oct; 35(10): 964-8. [DOI] [PubMed] [Google Scholar]
- 25.Kanellos I, Demetriades H, Zintzaras E, et al. Incidence and prognostic value of positive peritoneal cytology in colorectal cancer. Dis Colon Rectum. 2003 Apr; 46(4): 535-9. [DOI] [PubMed] [Google Scholar]
- 26.Leather AJ, Kocjan G, Savage F, et al. Detection of free malignant cells in the peritoneal cavity before and after resection of colorectal cancer. Dis Colon Rectum. 1994 Aug; 37(8): 814-9. [DOI] [PubMed] [Google Scholar]
- 27.Umpleby HC, Fermor B, Symes MO, et al. Viability of exfoliated colorectal carcinoma cells. Br J Surg. 1984 Sep; 71(9): 659-63. [DOI] [PubMed] [Google Scholar]