Need for adjuvant chemotherapy after colon cancer surgery – has it decreased?

Colorectal cancer (CRC) is the third most common cancer worldwide. Increased screening, early awareness, and improved treatment mainly explain decreasing mortality, despite rising incidence. One of the treatments for colon cancer contributing to improved survival is adjuvant chemotherapy. Its effect, at least in stage III, appears in many randomized trials [1]. Its effect is less well established in stage II, however, although is considered sufficient to motivate treatment when risk factors are present for recurrence. The decision as to the importance of risk factors varies between guidelines [2–4]. For stage II patients, tumour depth pT4 has been a predominant factor indicating higher risk for recurrence, but also perineural or vascular invasion, obstruction or perforation have frequently been associated with worse outcome, as is a low number of sampled lymph nodes. Microsatellite instability confers less recurrence risk [1,5].

In this issue of Acta Oncologica, Nozawa et al. [6] report pT4 to be the most important factor indicating higher risk for recurrence in stage II. In this Japanese study of 384 CRC patients treated in 2005–2014, approximately 10% received adjuvant chemotherapy. The study’s strength is their sufficient lymph node yield for the majority of patients (85%). Their reporting of possible loss of follow-up, confounders, patient selection and hence, generalizability, however, leaves questions unanswered. They present results for CRC patients, of whom 74% had colon cancer. In addition to pT4, preoperative carbohydrate antigen 19-9 (CA 19-9), which plays a significant role in pancreatic cancer, was of independent prognostic value. Preoperative carcinoembryonic antigen (CEA) did not influence prognosis.

In a preceding issue of Acta Oncologica, Hoshino et al. [7] similarly found the pronounced influence of infiltration depth. Their study created a nomogram to predict the probability of recurrence in stage II CRC, not separating colon and rectal cancer. Their nomogram had a bootstrapped concordance index of 0.64; 5-year recurrence-free survival (RFS) was 83%. For a male patient aged 64 with a pT3N0 tumour in the right colon, without perineural or vascular invasion, and with a preoperative CEA level of 5.0, and 15 investigated lymph nodes, the nomogram predicted 5-year freedom from recurrence at 93%. Interestingly, Weiser et al., 8 years earlier, created for stage I to III colon cancer another nomogram for the same fictive patient; it showed 5-year freedom from recurrence to be 85% [8]. Reasons for this difference are several. The Weiser et al. study based its results on a specialized single-centre institution in 1990–2000. Patients were mostly elective, and hence none had perforation or obstruction, which restricts generalization of its results. Both omitted data on emergency surgery, and on mismatch repair status [7,8].

In the two Japanese trials, risk for recurrence within the first 5 years for stage II patients was, overall, 13% [6] and 14% [7]. Adjuvant chemotherapy has influenced these seemingly low recurrence rates, but was given to few individuals, only 10% [6] and 25% [7]. This reflects a limited need for adjuvant therapy for all stage II patients today in Japan at centres having apparently high-quality surgery and pathology. In one systematic review aiming to describe recurrence risks in patients undergoing surgery during the past 10–15 years, with modern staging and surgery, 5-year disease-specific survival (DSS) for stage II colon cancer patients was about 83% [9]; the corresponding figure for stage III was 64%. Few studies, however, reported details on quality of care or on recurrence rates. Disease-free survival (DFS) and RFS are appropriate endpoints for clinical trials [10], when a need exists to balance gains (fewer recurrences) and negative consequences (toxicity including toxic deaths and second malignancies). These endpoints fail, however, to measure well the need for adjuvant therapy. Inclusion of endpoints other than recurrence, DFS (all deaths and second malignancies), and RFS (all deaths) overestimate that need. Cancer-specific survival (CSS) may underestimate the need for adjuvant therapy, since some recurrences are curable. Time to tumour recurrence (TTR), including recurrences of and death from the same cancer, seems the most appropriate measure, but was rarely reported [9].

Deciding on adjuvant therapy depends on differences between characteristics (like pT3 or pT4), and also on absolute risk for recurrence. Most studies exploring risk factors for recurrence do not report absolute risks. Knowledge of absolute risks facilitates dialogue with the patient as to whether the possible gain is sufficiently large to consider adjuvant treatment. Shared decision-making between doctor and patient, according to a Dutch study on adjuvant treatment of CRC, is rare [11,12]. Following trials showing gains from adjuvant therapy, major improvements have occurred in surgery, pathology, and staging. Absolute recurrence risk is presently lower, overall and at each stage (taking into account also stage migration), with the extent of improvement unclear [13–15]. The answer is yes to the question in our title, but how much has it decreased has no good answer.

Most stage II patients should not be offered adjuvant chemotherapy

Besides the two Japanese studies, several other recent studies indicate mostly a very low recurrence risk for colon cancer stage II. In a US study including 871 stage II patients, of whom 20% received adjuvant chemotherapy, 5-year RFS was 91% for patients with pT3 tumours; for those 7% with pT4 tumours, RFS was 72%, again emphasizing the importance of pT stage for recurrence risk [16]. In a South Korean study, pT4 stage, bowel obstruction or perforation, and high preoperative CEA indicated reduced RFS and DFS [17]. TTR went unreported, but patients who underwent surgery in 2006–2009 had an excellent RFS of 88% regardless of receiving adjuvant chemotherapy (preferably high-risk) or not, further showing recurrence risk to be small. Further, having no or only one risk factor meant a very low recurrence rate (6–7%; 18–50% with two or more risk factors), independent of adjuvant chemotherapy [18]. In that study, preoperative CEA was an independent risk factor for recurrence, unlike in one Japanese study [6]. In a German study, 492 colon cancer stage II patients treated with surgery only, no adjuvant chemotherapy, had a 5- and 10-year CSS of 93% and 88% [14]. A small subgroup (17%) with two or more risk factors had a 5-year CSS of 88%.

Do some stage II patients benefit from adjuvant chemotherapy?

In one study, only high-risk stage II patients, especially those with pT4 disease, seemed to benefit from adjuvant chemotherapy [19]. Due to harmful side-effects, low-risk patients who received adjuvant chemotherapy had worse survival. In Japan, adjuvant chemotherapy led to no improvement in RFS in high-risk patients [6], but a US registry study with over 153,000 stage II colon cancer patients having surgery in 1998–2011 showed that both low- and high-risk patients benefitted from adjuvant chemotherapy [20]. In this US study, choice of recipients of adjuvant chemotherapy was largely dependent on socioeconomic status and health care insurance, both introducing major bias. It is noteworthy that 5-year overall survival (OS) for all stage II patients was 81%, low-risk stage II patients without adjuvant chemotherapy had a 5-year OS of only 68%, whereas for those receiving it, OS was 86% [20]. One Italian study concentrating solely on pT3N0 reported that adjuvant-treated (AC) patients did much better than did those without it, with no major differences between low- and high-risk (RFS for AC low-risk 87% and high-risk 83%, versus no AC low-risk 75%, high-risk 71%) [21]. As their recurrence rate was markedly higher for those without adjuvant chemotherapy, these authors asked whether all stage II patients should be offered therapy. Uncontrolled studies seldom fully compensate for selection bias; their healthier patients receive adjuvant therapy and thus have better OS (and DFS and RFS), but not necessarily better TTR. When a EURECCA study compared the relative survival of stage II colon cancer patients in seven European countries, relative excess risk of death was lowest in Sweden, with no demonstrable relationship with adjuvant chemotherapy [22]. Sweden’s focus on colon cancer surgery during the past 10 years may explain these results [23].

Based on the Adjuvant Colon Cancer Endpoints (ACCENT) database of over 30,000 patients enrolled in trials starting in 1977 from over 40 countries, the risk for recurrence in stage II has been, after 5 years, 15% for those receiving only 5-fluorouracil (FU) and 10% with both FU and oxaliplatin [24]; as only 15% were stage II, only high-risk patients had likely been eligible. The ACCENT database highlights the potential of oxaliplatin in a high-risk stage II population. Adding oxaliplatin reduces recurrence risk by 33%, although trials only show an 18–20% risk reduction in DFS. In stage III, 37% had a recurrence after FU, again providing a potential gain from oxaliplatin, after which DFS decreased to 29%. The relevance of these recurrence risks after adjuvant FU alone is unclear. Trial patients are doing better, but above all, these patients had staging and surgery decades ago.

Improved surgery as a likely reason for less need of adjuvant chemotherapy

Although surgery is the backbone of colon cancer treatment, studies on quality of surgery are surprisingly few. Complete mesocolic excision (CME), together with high central vascular ligation, has improved survival based upon non-randomized evidence [25–28]. CME may have contributed to the nationwide improvement in survival recently evident in Denmark [29], although Quirke & West state the need for a large, international randomized trial of CME with or without central vascular ligation [30]. We agree, although for surgeons and centres who have already adopted the CME technique, a randomized approach may be difficult to initiate because the new method is considered superior. Initiating the CME technique necessitates both high-quality pathology and feedback on the specimen.

Can some stage III patients be spared from oxaliplatin-based chemotherapy?

Results from at least three randomized trials have demonstrated a benefit for stage III patients from adding oxaliplatin to a fluoropyrimidine [31–33], despite its increased toxicity, especially appearing as peripheral sensory neuropathy. A higher cumulative dose of the oxaliplatin is in fact, associated with more peripheral neuropathy; patients who required dose reduction due to acute neuropathy were at more risk for long-term neuropathy than were those tolerating a higher cumulative dose [34]. Based on a study in Sweden, neuropathy may differ depending on climate [35]. The summer group (treated in April to September) received a higher cumulative dose of oxaliplatin and tended to have more symptoms than did the winter group (October to March).

Since most stage III patients have received adjuvant chemotherapy, during recent years mostly given together with oxaliplatin, retrospective analyses evaluating outcomes in modern series of stage III patients are more difficult to interpret than are stage II series in which few patients received it. However, even after adjusting for adjuvant chemotherapy, each one-node increase in positive lymph node number elevated risk for recurrence by 24% [16], and stage IIIA patients clearly had fewer recurrences than did stage IIB patients [36]. The important question is whether some subgroups have such a low risk for recurrence that the extra benefit of oxaliplatin (18–20% relative reduction) is not justified.

Prognostic tools for risk stratification have not received the attention they deserve. The ACCENT group developed a web-based calculator for prediction of OS and TTR [37]. Based on 15,936 stage III patients they produced a tool with concordance indices of 0.66 for OS and 0.65 for TTR; the 7th edition of the American Joint Committee on Cancer (AJCC) staging system gave corresponding indices of 0.58 for OS and 0.55 for TTR. The ACCENT-based calculator was also superior to Numeracy [38], developed in 2004 for prognostication of stages II and III colon cancer. The fact that results from the ACCENT database are derived from phase III trials implies, however, selected patient inclusion. This limits the translation of results to a general population, but above all, these studies included patients treated decades ago, likely facing higher risks for recurrence than today.

Need for and benefit of chemotherapy for elderly patients

Few studies have concentrated on benefits of adjuvant chemotherapy in the elderly, and from randomized trials, they have even been excluded. In most countries, over half the patients are past 70 at initial diagnosis [39]. Their 1- and 5-year relative survival has improved, although these improvements are not as marked as in younger patients [40]. The proportion receiving adjuvant chemotherapy, and in particular oxaliplatin, has been much lower in the elderly – thus the low improvement figures [39]. It appears that more important than type of regimen is completion of all cycles of adjuvant therapy [41]. Patients over 70 without adjuvant post-operative chemotherapy had a 5-year RFS of 38% compared with 63% for CapMono-treated and 60% for CAPOX-treated patients. In general, patients receiving adjuvant chemotherapy are healthier, with fewer comorbid diseases. As RFS and OS may be similar irrespective of treatment with CapMono or CAPOX, choice of a less toxic treatment regimen is more likely to allow completion of all cycles. Elderly patients likely need adjuvant therapy as often as do younger patients, but they also need tailored treatment taking into account performance status, polypharmacy, and need for psychological and social support.

New prognostic tools still needed

Molecular-marker subtyping may provide new insights into need for adjuvant chemotherapy. Multiple studies have explored one or a few markers, but little practically important information has emerged [42–46]. Upon evaluation of the need for adjuvant chemotherapy, among molecular markers only microsatellite instability is presently recommended [47].

More promise lies in gene arrays. The 12-gene Recurrence Score assay (Oncotype DX Colon Recurrence Score assay; Genomic Health Inc., Redwood City, CA, USA) has been tested in stage II patients undergoing only surgery [48,49] and in stage III patients receiving adjuvant therapy [50]. Recently, a Japanese group validated the Recurrence Score in stage II and III patients treated with surgery only, noting that patients with stage II with a high-risk Score had a 5-year recurrence risk similar to those at stage IIIA and IIIB with low-risk Recurrence Scores (19% vs. 20%) [51]. Furthermore, high-risk stage IIIA-B patients had a recurrence risk similar to those with as low-risk stage IIIC (approximately 38%). These results remained in multivariable analyses, also confirming the importance of the 12-gene Score in prognostication, noting again that the recurrence risk for some high-risk stage II patients may be greater than for some individuals with stage IIIA disease.

Another promising biomarker is the caudal-type homeobox transcription factor 2 (CDX2) [52]. Especially among stage II patients, those CDX2-negative had a clearly shorter survival. However, CDX2-negative patients receiving adjuvant therapy did very well, indicating that CDX2 may be useful in selection for adjuvant treatment. Will this or other biomarkers, including colon subsite [53], add clinically important information as to when the risk of recurrence is so high that it is justified to lower it beyond the information provided by conventional, always recorded variables? Independent information beyond, for instance, pT-stage in stage II is insufficient. Can any test identify – among a group of patients having their recurrence risk by conventional risk factors as under 10% – any subgroup with a risk exceeding some 15% or 20%? Or the reverse? A test predicting whether a tumour (or actually disseminated subclinical tumour deposits) are sensitive to particular drugs would be much more valuable.

Neo-adjuvant rather than adjuvant chemotherapy

Neo-adjuvant chemotherapy is a possible alternative to adjuvant therapy. Two phase II trials, one randomized and both including colon cancer patients with T3 tumours with extramural tumour invasion more than 5 mm or with pT4 tumours, showed encouraging pathological responses in FOLFOX/XELOX-treated patients with or without panitumumab [54,55]. Those responding well to neo-adjuvant treatment also seemed to have a better DFS than did those requiring also postoperative chemotherapy. Both trials have continued as phase III. Neo-adjuvant therapy may provide an exciting alternative; results are eagerly awaited. Success relies on whether imaging or other tests can reliably identify groups with sufficiently high risks of recurrence to indicate pre-treatment.