Incidence of asymptomatic brain metastases in metastatic colorectal cancer

Abstract

Background:

Current literature suggests that brain metastasis is an infrequent occurrence in metastatic colorectal cancer. Outside of rare autopsy studies, these retrospective reports describe the incidence of symptomatic brain metastasis and therefore lack a description of the incidence in asymptomatic patients. With improved survival and a lack of routine brain imaging, the true incidence of brain metastasis among patients with metastatic colorectal cancer is likely underrecognized. At our research institution, protocol criteria require brain imaging regardless of neurologic symptoms. Therefore, we aim to describe the incidence of asymptomatic brain metastases in patients with metastatic colorectal cancer.

Patients:

Patients with metastatic colorectal cancer enrolled onto a clinical trial screening protocol at the National Cancer Institute that underwent brain imaging (n=171) between 2010–2019.

Results:

Median age of patients at initial colorectal cancer diagnosis was 48.1 years. Most had Stage IV disease with synchronous metastases. Twenty-five (14.6%) patients were identified with brain metastases, of which 19 (76%) were asymptomatic. Those with asymptomatic lesions were more likely to have presented with synchronous metastases, have a shorter time from primary diagnosis to development of metastatic disease, and have smaller brain metastases.

Conclusion:

We identified a high number of asymptomatic brain metastasis and subsequently a higher cumulative incidence of brain metastases in metastatic colorectal cancer patients than historical reports would suggest. This may represent a heretofore unknown aspect of the natural course of disease now being exposed due to an increasing life expectancy of these patients and could play a pivotal role in therapeutic decisions.

MICROABSTRACT:

Current literature reports that brain metastasis in patients with colorectal cancer has been a historically rare occurrence. Our study retrospectively analyzed 171 patients with metastatic colorectal cancer enrolled on a clinical trial screening protocol that included brain imaging and found a 14.6% incidence of which 76% were asymptomatic. These findings may have an important impact in treatment decisions for patients.

Background:

Colorectal cancer remains a leading cause of cancer and cancer-related deaths in the United States despite an improving 5-year overall survival rate over the last century due to the advent and evolution of screening initiatives, chemotherapy, metastasectomy, and other therapies.¹ The overall survival of patients with metastatic colorectal cancer has also increased from one year to 30 months with today’s combination chemotherapies.^2–5 The development of new strategies such as oligometastatic resection, monoclonal antibodies, targeted therapies, and immunotherapies have also demonstrated overall survival benefits, which can only be expected to improve as multimodality therapies evolve and new therapies are developed.^6–8

The occurrence of brain metastases in patients with colorectal cancer has been historically rare, with most literature citing an incidence of 1–3%.^{2, 9–19} Two studies, using autopsies as a method to describe the natural course of metastatic spread of colorectal cancer, report an incidence of brain metastases as 2.5–5%, but the application of these two studies are limited.^{11, 19} One predates modern chemotherapy—it was performed from 1944–1984—and the other did not perform brain autopsies on all reported patients, potentially skewing the true incidence of brain metastases reported. The remainder of the literature consists of multiple retrospective reviews that report varying incidences of brain metastasis, ranging from <1% to 8.8%.^{2, 9, 10, 12–18} A major limitation to these retrospective studies is their restriction of reporting only on symptomatic brain metastases as routine surveillance brain imaging was not performed.

Given the low reported incidence rates in the published literature, it is not surprising that current recommendations for metastatic evaluation do not include surveillance brain imaging except in cases where patients exhibit new neurologic symptoms. At our institution, we routinely perform imaging studies of the brain, regardless of symptoms, as required by clinical trial protocol criteria. This gives us a unique opportunity to evaluate the incidence of asymptomatic and symptomatic brain metastases. We hypothesize that the incidence of brain metastases is underreported.

Methods:

This is a single-center retrospective review of all patients with the diagnosis of metastatic colorectal cancer referred to the Surgery Branch of the National Cancer Institute (NCI) from 2010–2019 and enrolled into an IRB-approved protocol evaluating patients for clinical trial eligibility (NCT00001823). Patients did not proceed to brain imaging unless all other eligibility criteria were met, and those without brain imaging were excluded. Patients with known brain metastasis were not eligible. All patients provided informed consent prior to enrollment.

Demographic data and cancer-related medical history was obtained through medical charts. All patients included were diagnosed with colorectal cancer and colorectal metastases prior to referral to the NCI. Consequently, not all tumor characteristics and details regarding primary diagnosis (e.g.: stage, location) were available by medical records at our institution. If not clearly stated in available medical records, staging was assigned based on the 8^th edition of American Joint Committee on Cancer (AJCC) TNM staging criteria. Synchronous metastases were defined as any metastatic lesions diagnosed within six months of primary diagnosis of colorectal cancer. Metachronous metastases were those diagnosed after 6 months of colorectal cancer diagnosis.

All brain metastases were diagnosed solely on radiographic imaging studies consisting of magnetic resonance imaging (MRI) or, less frequently, computed tomography (CT) scans. All imaging studies were read by a neuroradiologist at our institution. The number and size of brain metastases were taken from official radiology reports. For statistical analysis, the diameter of the largest brain metastases for each patient was used.

Our primary endpoint was to identify the incidence of asymptomatic brain metastases in patients with treatment-refractory metastatic colorectal cancer. Secondary endpoints included the identification of overall brain metastases incidence in patients with metastatic colorectal cancer as well as factors that may differentiate asymptomatic brain metastases vs symptomatic brain metastases.

Statistical Analysis:

All statistical analyses were performed using GraphPad Prism 7.01© (San Diego, CA). A Mann-Whitney U test was performed on nonparametric data. Fisher’s exact and chi-square tests were used to evaluate potential associations of developing brain metastases. When evaluating the impact of stage, unknowns were excluded and the Cochran-Armitage test for trend was applied. A log-rank test was applied to Kaplan-Meier curves to analyze overall survival. Patients lost to follow-up were censored as alive at date of last contact if date of death could not be identified by public records. Survival analysis was performed in April 2020. All values were considered statistically significant if the p-value ≤ 0.05.

Results:

Three hundred thirty metastatic colorectal cancer patients were enrolled onto protocol from 2010 to 2019 (Figure 1). Due to a lack of brain imaging, 159 patients were excluded. Of the remaining 171 patients, 162 had at least one brain MRI performed and nine patients who had at least one brain CT scans performed. Forty-six patients were diagnosed with rectal cancer while 125 patients were diagnosed with colon cancer. The median age at the time of initial colorectal cancer diagnosis was 48.1 years (95% CI 46.8–49.3), and most presented with Stage IV disease with synchronous metastases. Detailed patient demographics are displayed in Table 1. All patients had disease refractory to standard chemotherapy regimens including a 5-fluorouracil based therapy. Most patients had disease refractory to second- and third-line therapies prior to enrollment onto protocol. All patients had at least one site of metastatic disease at time of enrollment onto protocol with most having two to four sites of metastatic disease. The majority of metastatic sites included but were not limited to lung, liver, lymph nodes, and peritoneum. No patient had intracranial disease only.

Figure 1:

Schematic of retrospective review of patients with colorectal cancer screened by the Immunotherapy Referral Center of the Surgery Branch, National Cancer Institute, National Institutes of Health

Table 1:

Patient Demographics. Age and survival are reported as median (range).

	All Patients (171)	No Brain Metastasis (146)	Brain Metastasis (25)	p Value	Symptomatic Brain Metastasis (6)	Asymptomatic Brain Metastasis (19)	p Value
Sex
Male	81	71	10	0.52	2	8	>0.99
Female	90	75	15		4	11
Age at Diagnosis (years)	48.1 (26.4–70.1)	47.8 (26.4–70.1)	50.1 (35.0–62.6)	0.18	54.4 (42.8–62.6)	50.0 (35.0–58.2)	0.18
Primary
Colon	125	107	18	>0.99	4	14	>0.99
Rectal	46	39	7		2	5
Stage at Diagnosis
1	3	3	0	0.72	0	0	0.06
2	15	13	2		1	1
3	41	32	9		4	5
4	97	85	12		1	11
Unknown	15	13	2		0	2
Synchronous Metastatic Disease	104	89	15	>0.99	1	14	0.02
Metachronous Metastatic Disease	67	57	10		5	5
Time from Diagnosis to Metastatic Diagnosis (months)	0.5 (0–98.9)	0.5 (0–98.9)	1.5 (0.5–25.8)	0.45	11.6 (0.5–25.3)	0.5 (0.5–18.7)	0.05
Time from Diagnosis to Diagnosis of Brain Met (months)	-	-	42.8 (9.5–119.2)	-	53.7 (23.9–97.2)	41.3 (9.5–119.2)	0.51
Time from Diagnosis of Metastatic Disease to Diagnosis of Brain Met (months)	-	-	40.3 (9.1–118.7)	-	43.7 (12.3–71.4)	37.4 (9.1–118.7)	0.74
Survival (months)
From initial diagnosis	52.5 (8.6–219.6)	51.6 (8.6–219.6)	57.0 (13.5–130.8)	0.51	60.2 (28.1–111.4)	57.0 (13.5–130.8)	0.91
From detection of brain metastasis	-	-	-	-	5.8 (2.3–15.5)	3.3 (0.07–32.3)	0.78

Overall, 25 patients were identified with brain metastases, for a cumulative incidence of 14.6%. At the time of brain metastasis diagnosis, 19 patients (76%) were asymptomatic. The remaining six patients (24%) demonstrated neurologic symptoms consisting of headache, dizziness, altered mental status, aphasia, extremity weakness and paresthesia, or seizure which warranted clinical evaluation and led to the subsequent diagnosis of intracranial metastases.

Two patterns of screening emerged: patients with a single instance of brain imaging (n=93) and patients with multiple time points of brain imaging dictated by further clinical trial care (n=78). Of those patients with asymptomatic brain metastases, 10 were identified with disease on initial imaging for a presentation incidence of 5.8% (10/171). The remaining nine patients with asymptomatic brain metastases and all six patients with symptomatic brain metastases were diagnosed during a surveillance period for a surveillance incidence of 19.2% (15/78). Median time to development of brain metastases in these 15 patients was 9.2 months (range 1.3–24.0) from initial imaging (Figure 2).

Figure 2.

Pattern of imaging and detection of brain metastases in patients with metastatic colorectal cancer. Each red bar represents a single patient with a symptomatic brain metastasis (n=6), and each blue bar represents a single patient with an asymptomatic brain metastasis (n=19); grey bars represent the median values for patients without brain metastases. The top half of the chart describes those patients with negative initial scans; the lower portion details those patients with brain metastasis identified on initial imaging. Endpoints are death or last known contact, unless otherwise noted.

There was no significant difference in the incidence of brain metastases when evaluated by sex (12% male vs 17% female, p=0.52), age at primary diagnosis of colorectal cancer (median: 47.8 vs 50.1 years, p=0.18), site of primary (14% colon vs 15% rectal, p>0.99), stage at primary diagnosis (0%, 13%, 22%, 12%, stages 1–4 respectively and unknown 13%, p=0.72), timing of metastatic disease (14% synchronous vs 15% metachronous, p>0.99), or time from primary diagnosis to diagnosis of metastatic colorectal cancer (median: 0.5 vs 1.5 months, p=0.45).

When comparing patients with symptomatic brain metastases to those with asymptomatic brain metastases, two interrelated factors were identified as statistically significant: patients with asymptomatic brain metastases were more likely to have synchronous metastatic colorectal cancer at presentation (93%, p=0.02) and therefore, a shorter time from diagnosis of colorectal cancer to diagnosis of metastatic colorectal cancer (median: 0.5 vs 11.6 months, p=0.05) There was no significant difference between patients with symptomatic and asymptomatic brain metastases in the other potential associations in the development of intracranial disease (Table 1). The number and size of brain metastases were also compared across groups. Patients with symptomatic brain metastases were found with larger lesions comparatively (median: 2.75 vs 0.8 cm, p=0.0005), but there was no difference in the number of brain metastases (median: 1 vs 1, p>0.99) identified in each group. A total of eight patients had multiple brain metastases, six of which were asymptomatic at the time of diagnosis of intracranial disease.

No overall survival difference was found when comparing patients with and without brain metastases (median: 57.0 vs 51.6 months, p=0.51). To identify if there was an overall survival difference between patients with asymptomatic and symptomatic brain metastases, these cohorts were compared in two ways. First, overall survival was analyzed from the time of initial diagnosis of colorectal cancer and no difference was identified (median: 57.0 vs 60.2 months, p=0.91). Overall survival was also analyzed from the time of diagnosis of brain metastases in asymptomatic and symptomatic patients and, again, no significant difference was identified (median: 3.3 vs 5.8 months, p=0.78) (Figure 3).

Figure 3:

No significant difference in overall survival from initial diagnosis of (A) patients with or without brain metastasis (median 57.0 vs 51.6 months, p=0.51) or (B) patients with asymptomatic compared to symptomatic brain metastasis (median 57.0 vs 60.2 months, p=0.91). No significant difference in overall survival from time of diagnosis of brain metastasis (C) in patients with asymptomatic vs. symptomatic tumors (median 3.3 vs 5.8 months, p=0.78).

Discussion:

Current literature is largely restricted to the reporting of only symptomatic brain metastases in patients with metastatic colorectal cancer by retrospective reviews and two limited autopsy studies which do not accurately capture the true and undiagnosed incidence of brain metastases in today’s patient population. Here we present a retrospective review of 171 patients from 2010–2019 with metastatic colorectal cancer who have undergone screening for intracranial metastatic disease as dictated by eligibility criteria for various clinical trials at our institution. This unique review differs from current literature, as it evaluates potential intracranial metastatic disease in all patients with metastatic colorectal cancer regardless of neurologic symptomatology. In our analysis, we identified 25 patients with brain metastases, of which 76% were identified with brain imaging in asymptomatic patients. The remaining patients demonstrated various neurologic symptoms that warranted clinical evaluation which subsequently led to their diagnosis of intracranial disease.

Our reported cumulative incidence of brain metastasis in metastatic colorectal cancer patients is higher than both the highest reported incidence of 8% and the most commonly reported incidence rate of 1–3%.^{2, 9–19} However, this high cumulative incidence follows a similar trend to that reported by Christensen et al. in 2017 which reports the highest current literature incidence of 8% and has a similar cohort to ours. In addition, our identified incidence of symptomatic brain metastasis (3.5%) is consistent with historical reports and the current literature which is limited to reporting symptomatic lesions only given their retrospective nature. This suggests that although our cohort consists of a highly selected group of patients seeking experimental protocols, they continue to exhibit the known natural history course of the disease as expected in patients with metastatic colorectal cancer. Our identification of a high incidence of asymptomatic brain metastases, which accounts for most of the cumulative incidence of brain metastases in this study, gives rise to the concept that this may be uncovering a new facet of the natural history of the disease. This extension of the natural progression of metastatic disease is now possible given today’s more effective systemic therapies and perhaps indicates that intracranial pathology may be underrepresented in historical data.

Given the low historical incidence of brain metastases in patients with metastatic colorectal cancer, it is not surprising that current management guidelines for patients with colorectal cancer do not recommend evaluation or surveillance of the brain for metastatic disease. However, specific guidelines regarding brain imaging do exist for staging and surveillance of lung cancer and melanoma. A recent study evaluating the lifetime occurrence of brain metastases in lung cancer, breast cancer, and melanoma found a 13.5%, 1.8%, and 3.6% incidences, respectively.²⁰ Given our identified 14.6% cumulative and 11.1% asymptomatic brain metastases incidences are comparable or higher than these surveilled cancer histologies suggests that there may be a role for surveillance imaging in patients with metastatic colorectal cancer. Our data suggest that patients with metastatic colorectal cancer who have progressed through multiple lines of therapies are prime candidates for both an evaluation of metastatic brain disease as well as routine interval surveillance. Over half of the patients with brain metastasis in our cohort were identified on initial imaging and another 19% of those surveilled were eventually found to have intracranial disease.

Although our data did not show an overall survival difference between any of the cohorts, by performing routine surveillance brain imaging we were able to recognize silent brain metastases which were significantly smaller in size compared to those lesions causing neurologic symptoms. This is important for clinicians creating management plans for patients with refractory colorectal cancer, including those patients who are undergoing experimental treatment strategies. Identifying hidden brain metastases with surveillance brain imaging, smaller lesions are diagnosed which may have implications for treatment decisions both locally (e.g. stereotactic radiation vs resection) and systemically (e.g. metastasectomy of visceral metastasis vs systemic chemotherapy). The identification of brain metastasis will also invariably alter discussions of the risk profile and therapeutic benefits of potential future treatments. These can be key factors for physicians determining appropriate treatment modalities and for patients making both therapeutic and quality of life decisions.

Management of brain metastases is an evolving field with nuances beyond the scope of this analysis. Individual consideration should include the burden and pace of extracranial disease, the presence or absence of symptoms, control of the primary tumor, and underlying issues of quality of life. In a randomized trial of patients with 1–3 brain metastases and otherwise stable disease, the European Organisation for Research and Treatment of Cancer (EORTC) evaluated the addition of whole brain radiotherapy (WBRT) after complete surgical resection or stereotactic radiosurgery (SRS). The addition of WBRT reduced the relapse rate at the initial site and at new intracranial sites, irrespective of the initial approach, when compared to observation alone, however, did not improve duration of functional independence nor overall survival.²¹ Additionally, patients in the observation arms reported better health-related quality of life scores.²² The use of post-operative SRS has been explored as an alternative to adjuvant WBRT with less decline in cognitive function and similar overall survival.²³ When a larger burden of brain disease is encountered, WBRT is the most frequently applied treatment, but supportive care alone may be appropriate in the face of progressing disease and poor performance status.²⁴

In our analysis, we attempted to identify factors that may predispose patients to developing metastatic brain lesions. We learned that even though factors were not significant when compared on whole, when analyzed for differentiation between asymptomatic and symptomatic brain metastases, those with asymptomatic brain metastases more commonly presented with synchronous metastatic colorectal cancer at primary diagnosis and subsequently had a shorter time from primary diagnosis to diagnosis of metastatic colorectal cancer. A potential explanation for this finding is beyond the scope of this report. One hypothesis is that patients with asymptomatic disease have cancer cells that evolve early to establish residence in a neural pre-metastatic niche but are slow growing, whereas those with larger, symptomatic brain metastases may have clones with an aggressive growth phenotype.²⁵ However, there was no difference in overall survival between cohorts – likely an inability to detect small differences in survival with a small cohort.

One of the biggest limitations to our study was the retrospective nature of our single institution prospectively accrued database. The sampled patient population is a select group of patients being evaluated for clinical trials at the NCI. These patients remain in good performance status (ECOG 0 or 1), irrespective of advanced refractory metastatic colorectal cancer. In the closest comparison, those patients in our cohort that presented with metastatic disease demonstrate a slightly improved five-year overall survival (33%, 95% CI 22.2–44.1) when compared to a national database (14.3%).²⁶ However, the incidence of symptomatic brain metastases is similar to current literature. Due to strict eligibility criteria for our clinical trials, patients with known brain metastasis were not eligible, and therefore would not be included in our patient population. This would include those patients who present with neurologic defects as their presenting symptom leading to a diagnosis of colorectal cancer, though this number is likely small. Another limitation is the lack of longitudinal surveillance in those patients deemed ineligible for treatment protocols which hampers the ability to continue further evaluations for brain metastasis and survival data.

Conclusions:

With today’s advances in the treatment of colorectal cancer, patients with metastatic disease are living longer. Historically the incidence of brain metastases in patients with colorectal cancer has been rare as the identification of these lesions are limited to symptomatic patients. Our study identified a 14.6% cumulative incidence of brain metastases with 76% of those in asymptomatic patients. Comparable to melanoma and lung cancer, these results illustrate there may be a role for surveillance brain imaging in these patients. Surveillance imaging that identifies hidden brain metastases would likely inform therapeutic options and risk-benefit discussions between multidisciplinary care teams and their patients.

CLINICAL PRACTICE POINTS:

Based on the current literature, brain metastasis is a rare occurrence in patients with metastatic colorectal cancer. Most literature report an incidence of 1–3% but rare reports demonstrate an incidence upwards to 8%. The treatment landscape of colorectal cancer has changed drastically, and the survival profile of patients has improved. Despite these positive changes, there remain no standard guidelines to brain imaging. We hypothesized that the incidence of brain metastases largely goes unreported. In an effort to address this, we retrospectively reviewed 171 patients with metastatic colorectal cancer prospectively enrolled on a clinical trial screening protocol at the National Institutes of Health who underwent brain imaging. We identified 25 patients (14.6%) with brain metastasis. Of these, 76% (19/25) were identified in asymptomatic patients who outside of the protocol would not have had an indication for imaging otherwise. When comparing patients identified with brain metastasis, those with asymptomatic lesions were found to be smaller than those with symptomatic lesions. If the presence or absence of intracranial disease informs surgical and medical management, these findings may play a pivotal role in therapeutic decisions both for the treating physician and patient with treatment-refractory metastatic cancer.