Methods

A prospective study was performed on 72 colorectal resection specimens, removed for carcinoma, taken over a 16 month period (June 1998 to September 1999) in one teaching hospital. The specimens were dissected by two pathologists, QvW and MB. Specimen dissection and tissue sampling of the primary tumour was in accordance with the current guidelines for the histopathological assessment of colorectal carcinoma.^2,3 The sampling of retrieved lymph nodes was based on the current guidelines for the detection of breast cancer metastases in axillary lymph nodes as follows.⁸ Lymph nodes were detected by palpation and slicing of associated adipose tissue under a good light. The apical (highest) node was labelled separately when identified. Macroscopically identifiable lymph nodes up to approximately 5 mm in maximum extent (“small” nodes) were, without further sectioning, submitted in entirety and examined at three levels, at least 100 μm apart. Multiple small nodes were processed in one cassette. Nodes larger than approximately 5 mm in maximum extent (“large” nodes) were divided into at least two sections, up to the maximum number that could be processed in one cassette and examined microscopically at one level. Only one representative section was taken from lymph nodes that were macroscopically replaced by tumour. Particular attention was paid to matching adjacent slices of any nodes transected while slicing the bowel wall for the assessment of the primary tumour. After standard processing, sections for microscopy were cut at 4 μm thickness and examined for metastases using a standard haematoxylin and eosin stain only. Further small nodes detected only on microscopy of bowel wall blocks were subsequently examined at two further levels.

The number of lymph nodes containing metastatic tumour was recorded. For involved large nodes, metastatic tumour in any node section other than the larger/largest in a given cassette was regarded as a metastasis that might have gone undetected if only one section of the node had been processed (“additional” metastasis). For small nodes examined at three levels, any metastasis in levels two or three, but not in level one was regarded as a metastasis that might have gone undetected if only one level had been examined (“additional” metastasis). The numbers of involved and uninvolved, small and large nodes were assessed and staging by modified Dukes's stage and TNM classification was applied.^9,10 The assessment of additional metastases allowed us to determine whether the detection of these metastases led to upstaging of the tumour. To determine the extra workload generated by this method, the extra number of tissue blocks and slides generated for each case was estimated by counting the number of blocks and slides produced and subtracting the estimated number that would have been produced if single sections from multiple nodes had been processed together in one cassette.

Results

In total, 72 colorectal carcinoma resection specimens were assessed, including eight Dukes's A, 26 Dukes's B, 29 Dukes's C1, and nine Dukes's C2. The mean and median numbers of lymph nodes identified for each case were 13 and 12, respectively (range, three to 44). The mean numbers of lymph nodes from Dukes's A, B, and C cases were seven, 13, and 15, respectively (medians, 11, 11, and 12, respectively). In 52 cases, 10 or more lymph nodes were identified. The single case from which only three lymph nodes were retrieved was Dukes's stage C. For the year before the start of our study the frequency of Dukes's A, B, and C in resection specimens recorded at our hospital was 12%, 47%, and 40%, respectively, and the mean number of nodes retrieved was six, 11, and 10, respectively. The mean number of nodes for all cases was 10 (range, 0–28).

In five lymph nodes from four patients, metastatic deposits were identified that might have gone undetected by the widely accepted method of assessment. In four large lymph nodes from three patients, metastases were identified in only one of the multiple sections of a single node and did not involve the largest slice from the node. In one further case an additional nodal metastasis was found on examining levels of a small node; a metastatic deposit was present in only the second and third levels of the lymph node. However, in only one of the four patients did the detection of an additional metastasis lead to upstaging of the tumour from Dukes's B to C and from N0 to N1 with regard to TNM classification. This case showed a lymph node metastasis in the smallest of three slices of a large node and was the only metastatic deposit identified in a total of 44 nodes. The lymph node measured at least 7 mm in maximum extent and the metastasis was present in the smallest of the three slices taken through that node. In the plane examined, this metastasis measured 2.7 mm in maximum extent. The three other cases described above showed widespread metastases in further nodes; therefore, the additional metastases in those cases did not alter the tumour stage.

With regard to the additional workload generated by this method, an estimated 414 extra tissue blocks were processed (mean, six/case). The approximate number of extra slides was greater (535) because blocks of small nodes had three levels cut, requiring further slides (table 1).

Discussion

In our study, five lymph node metastases identified from four colorectal carcinoma resection specimens might have gone undetected if the current widely applied method of lymph node sampling had been used. Although the number of additional metastases detected was small, in one case it led to upstaging from Dukes's B to C. It is not possible to confirm with certainty whether or not the additional metastases detected would have gone undetected if only one section of lymph node was processed, because this would depend on the plane of sectioning by the pathologist. When choosing which piece of a sectioned lymph node to process for microscopic examination, there is obviously an element of chance involved as to whether a metastatic deposit replacing less than half of the lymph node volume will be detected. It has been proved mathematically that the chance of identifying a randomly distributed 1 mm lesion in a 5 mm node that is bisected is as low as 37%.⁷

It has been shown that, with regard to colorectal cancer resection specimens, the greater the number of lymph nodes examined, the greater the chances are of detecting metastases in these nodes.¹ Although it was not the aim of our study to investigate this factor, it is interesting to note that, of the cases studied, 53% were Dukes's stage C and the mean number of lymph nodes retrieved for each case for this stage was 15, whereas in the year before our study the figures were 40% and 10, respectively. Methods have been applied previously to increase the yield of lymph nodes. Scott et al used fat clearance (by means of xylene and alcohol) to increase the number of lymph nodes retrieved from colorectal carcinoma resection specimens.¹¹ The five year follow up of patients to whom this method was applied found it to be of clinical relevance.¹² Fat clearance is relatively inexpensive but is regarded as impractical and unsafe. The health and safety aspect of this technique is the major reason that it has not gained wide acceptance. Both xylene and alcohol are highly inflammable agents and certain occupational exposure limits are laid down for these and other chemicals used in laboratories.¹³ Methods have also been described that increase the detection of occult metastases in lymph nodes, including immunohistochemical staining for cytokeratins and molecular biological techniques.^4,5,14,15 Despite the increased detection of metastases by these methods, they have not been widely accepted as routine practice.

Our study is based on the fact that careful macroscopic specimen assessment still appears to be the most widely accepted means of optimum lymph node retrieval in the UK.

Our results suggest that the number of important (with regard to tumour stage) additional metastatic deposits detected by our method, compared with the usual method of lymph node sampling, appear to be small.

If our method was applied routinely, the time and therefore cost of colorectal carcinoma reporting would be increased for both laboratory and medical staff, although the accuracy of tumour staging might be improved. Given current financial and manpower restraints, it is unlikely that such a method would be widely regarded as appropriate for routine application in view of the low yield of additional metastases. However, pathologists should be aware of the possibility of “missed” metastases and should, at least, select sections from lymph nodes in such a way as to allow the largest possible cut surface to be assessed—for example, by sectioning ellipsoid shaped nodes along their long axes. Lymph node detection is, after all, a time consuming part of specimen assessment and, once identified, each node should be handled in the optimum way for accurate staging. We cannot assess the clinical relevance of the additional metastases detected in our study; however, we have given an indication of the possible extent of metastases missed by processing partial lymph nodes only. Pathologists involved in compiling guidelines for reporting of cancer resection specimens should address the handling of lymph nodes with a view to achieving a uniform approach.