Extended Clavien-Dindo classification of surgical complications: Japan Clinical Oncology Group postoperative complications criteria

Purpose Prior to publication of the Clavien-Dindo classification in 2004, there were no grading definitions for surgical complications in either clinical practice or surgical trials. This report establishes supplementary criteria for this classification to standardize the evaluation of postoperative complications in clinical trials. Methods The Japan Clinical Oncology Group (JCOG) commissioned a committee. Members from nine surgical study groups (gastric, esophageal, colorectal, lung, breast, gynecologic, urologic, bone and soft tissue, and brain) specified postoperative complications experienced commonly in their fields and defined more detailed grading criteria for each complication in accordance with the general grading rules of the Clavien-Dindo classification. Results We listed 72 surgical complications experienced commonly in surgical trials, focusing on 17 gastroenterologic complications, 13 infectious complications, six thoracic complications, and several other complications. The grading criteria were defined simply and were optimized for surgical complications. Conclusions The JCOG postoperative complications criteria (JCOG PC criteria) aim to standardize the terms used to define adverse events (AEs) and provide detailed grading guidelines based on the Clavien-Dindo classification. We believe that the JCOG PC criteria will allow for more precise comparisons of the frequency of postoperative complications among trials across many different surgical fields.


Introduction
The evaluation of postoperative complications in surgical trials is as important as the assessment of toxicities in chemotherapy trials. Prior to the proposal of a therapy-oriented classification scheme, by Clavien PA et al. in 1992 [1], there were no accepted definitions for the grading of surgical complications in clinical practice. This framework proposed by Clavien et al. was not used widely, because there was no system for the grading of severity of surgical complications [2] and no uniform definition of these events. For instance, some surgeons included a body temperature greater than 38 °C on two consecutive days as being "high", whereas others included intraoperative complications, postoperative complications (within 30 days), and late events such as dumping syndrome. Few randomized controlled trials (RCTs) [3] have used this classification system, with individual parochial definitions of surgical complications being used in most surgical RCTs [4][5][6].
In cancer clinical trials, adverse events (AEs) are evaluated in accordance with the Common Terminology Criteria for Adverse Events (CTCAE), which is far from exhaustive in terms of surgical complications; thus, some surgeons are not comfortable using grading definitions. The Clavien-Dindo classification, published in 2004 [7] defined a simple classification of postoperative complications, which has been adopted widely in clinical practice. Although this classification categorizes postoperative complications broadly into four major groups, it is often desirable to more clearly define the common AEs to avoid the use of different or less precise terms for the same AEs occurring in different clinical trials. More detailed grading criteria for common AEs would also be helpful for surgeons. Therefore, our aim was to establish supplementary criteria for the Clavien-Dindo classification to standardize the evaluation of postoperative complications.

Methods
The Japan Clinical Oncology Group (JCOG) commissioned a committee to establish more precise criteria for the grading of surgical complications. The committee comprised members from nine JCOG study groups (gastric, esophageal, colorectal, lung, breast, gynecologic, urologic, bone and soft tissue, and brain) who have extensive experience with surgical trials. These groups established the JCOG postoperative complications criteria (JCOG PC criteria). Members identified the postoperative complications experienced commonly in their fields and defined detailed grades for each complication in accordance with the general grading rules of the Clavien-Dindo classification. The JCOG PC criteria were reviewed and approved by the JCOG Executive Committee and published on the JCOG website in October, 2011 (in Japanese) [8].

Results
The JCOG PC criteria included 72 surgical AEs experienced commonly in surgical trials, including 17 gastroenterological complications, 13 infectious complications, six thoracic complications, and several other complications (Table 1). If no applicable AE terms are found in the JCOG PC criteria, 'other (specify)' should be chosen. In such cases, the appropriate AE term should be used, and the overall grading should be performed in accordance with the general rules of the Clavien-Dindo classification. Because the grading definitions follow the general rules of the Clavien-Dindo classification, surgeons can use these original rules to grade AEs, and can also refer to the more detailed definitions in the JCOG PC criteria if necessary. Table 2 lists the differences between CTCAE, the Clavien-Dindo classification, and the JCOG PC criteria.

Discussion
Until Clavien PA et al. published their original classification in 1992, there were no established criteria or framework available to standardize surgical complications in surgical trials. In 2003, the US National Cancer Institute-Common Toxicity Criteria (NCI-CTC) version 2.0 [9] were revised and renamed the CTCAE version 3.0 [10]. This system has been used widely to evaluate and define the toxicity of chemotherapy or radiotherapy. While terms and definitions for AEs occurring as a result of intraoperative and postoperative complications were not included in the NCI-CTC version 2.0, some surgical AE terms were incorporated in the CTCAE version 3.0. Nevertheless, the CTCAE version 3.0 failed to include many surgical complications and surgeons were frequently unable to objectively classify complications using its grading definitions.
In 2009, the CTCAE version 4.0 [11] was released, with considerably more surgical AE terms, but several common surgical complications were still not included. For example, intra-abdominal abscess, pyothorax, delayed gastric emptying, and lung torsion were not listed as AE terms. Moreover, grading definitions were not clinically optimized for some surgical AEs. For example, the grading definition of pancreatic fistula in this version of the CTCAE is suitable for pancreatitis, but not for pancreatic fistula after pancreatectomy. Such inappropriate definitions have made surgeons reluctant to use the CTCAE version 4.0 in surgical trials.
In 2004, the Clavien-Dindo classification was modified to allow for the grading of life-threatening complications and long-term disability caused by a complication. This revised version defines five grades of severity (Grade I, II, IIIa, IIIb, IVa, IVb, and V) and the suffix "d" (for "disability") is used to denote any postoperative impairment [7]. This refined Clavien-Dindo classification has been used increasingly in clinical practice and also in clinical trials involving surgical procedures, because it is simple, reproducible, and flexible [12]. Rather than providing specific grading criteria for each AE, the Clavien-Dindo classification provides broad-based but general criteria that can be used uniformly for all kinds of surgical AEs. However, several issues have emerged since this classification became more widely used. One controversial issue is that AE terms are not well defined and different AE terms designate the same AEs in different clinical trials. For example, when intestinal obstruction occurs, some investigators could report this AE as "ileus", but others refer to it as "small bowel obstruction" or "colon obstruction". Under such circumstances, the incidence of this AE cannot be counted accurately. A second issue is that only general grading criteria are defined and therefore, grading can be difficult in some cases and subject to bias by the grader. For example, primary non-operative treatment for intestinal obstruction is gastroenteric tube decompression. Nasogastric tube or nasoenteric tube is utilized depending on the severity, but the original Clavien-Dindo classification does not define what grading should be applied for any type of gastroenteric tube placement for decompression. The JCOG PC criteria were established to address these issues. The advantages of the JCOG PC criteria are as follows: First, commonly experienced surgical AEs are specified and listed. To compare precisely the frequency of surgical complications between studies, use of the common AE terms specified in the JCOG PC criteria is recommended. Second, grading definitions are straightforward and optimized for surgical complications. With these advantages, the JCOG recommends that the JCOG PC criteria be used to supplement the Clavien-Dindo classification, while maintaining the overall Clavien-Dindo classification. In JCOG, some disease-oriented subgroups are conducting clinical trials including surgery and using both the CTCAE and JCOG PC criteria to evaluate postoperative complications. After these trials are completed, we will evaluate the concordance between the grading by the CTCAE and that by the JCOG PC criteria. We also plan to explore the advantages and disadvantages of the JCOG PC criteria.
The JCOG PC criteria have some limitations. First, these AE terms were chosen somewhat arbitrarily, but by experienced surgeons, and specific grading was decided based on the opinions and experience of our committee members. A second limitation of the JCOG PC criteria is that they do not include intraoperative complications. Our intent was to further define and clarify the criteria of the Clavien-Dindo classification and we considered that incorporating intraoperative complications would deviate too much from the original concept. Another common classification may be required to define and grade intraoperative complications. A third limitation is that all descriptions in the Clavien-Dindo classification pertain to early postoperative complications. Here, 'early postoperative' generally indicates the time from surgery to the first hospital discharge, but in theory, the Clavien-Dindo classification can be applied broadly to late postoperative complications after hospital discharge. Within this context, the JCOG PC criteria are mainly intended to be used for early postoperative complications, but they can also be used after hospital discharge, although would require more definitions and AEs.
In conclusion, the goals of the JCOG PC criteria are to standardize the AE terms used for early postoperative complications by providing more detailed grading guidelines based on the Clavien-Dindo classification. We suggest that researchers use the JCOG PC criteria in every surgical trial to allow for precise comparison of the frequency of surgical complications among trials.