Adrenal metastases in lung cancer: Clinical implications of a mathematical model

Abstract

Adrenal gland metastases are common in lung cancer. It is well recognized that aggressive treatment of solitary adrenal metastases leads to improved outcomes but the exact nature of adrenal deposits is not well understood. Controversy exists as to the routing of cancer cells to the adrenal gland with some believing that this transmission is lymphatic, in contrast to the more generally accepted theory of hematogenous spread. Recently published mathematical modeling of cancer progression strongly supports the lymphatic theory. With that in mind we performed a literature review to look for biological plausibility of simulation results and believe that evidence supports the contention that metastases to the adrenal gland can be routed via lymphatic channels. This could explain improved survival for patients in whom solitary adrenal metastases are managed aggressively with surgical or radiation modalities. We are calling for clinical trials prospectively testing this hypothesis.

Introduction

For over a century, metastases have been acknowledged to be unpredictable. Charles Powell White observed in his book Tumors in 1913 that “We may have a carcinoma with extensive involvement of the lymph glands and no visceral deposits, while we may find another similar carcinoma with extensive visceral tumors and no affection of the glands; or, again, we may find an extensive primary tumour without involvement of either glands or viscera”. [1] Despite advancements in medical knowledge, we have failed to quantify our understanding of metastatic progression patterns. One example is the controversy in primary lung cancer and adrenal metastases. In lung cancer, although adrenal metastases are common, there is no clear consensus as to whether they occur via a lymphatic or hematogenous route. Perhaps because lung anatomy does not clearly reinforce connections of lymphatics between lungs and adrenals, many would regard these metastases as primarily hematogenous. Although the lymphatic theory of adrenal gland spread is not novel as it was described by Onuigbo in 1957, to date this finding is not commonly acknowledged in clinical practice. [2] A recently published Markov chain based mathematical model of metastatic progression of primary lung cancer supports the lymphatic pathway of spread.[3, 4] In this manuscript we briefly review the literature, discuss the new mathematical model of metastasis, and examine the strength of the evidence pointing to lymphatic connections between primary lung tumor and the adrenal gland. Proven correct this would result in a down staging of this patient population to Stage III and the related modification in treatment plan for curative intent.

Markov Chain Model’s confirmation of lymphatic adrenal spread

The mathematical model[3, 4], which we refer to as the ‘Newton’ model, uses probabilistic methods such as Markov chain dynamics, random walkers, and Monte Carlo simulations to simulate how a tumor cell moves from the primary tumor to each metastatic site. The model was generated by using an autopsy data set of 3827 untreated cancer victims. [5] For this report we will focus on lung cancer as a primary site. The model regards metastasis-inducing cells as random walkers on an anatomic network, each anatomic site serves as a node between which cells can travel to spread disease, the autopsy data set is treated as the steady state distribution of metastasis, and computational strategies are used to identify the route and mean first-passage times (MFPT) by which cancer cells spread to arrive at the steady state. MFPT is a quantity that tracks how many steps on average a random walker takes to reach a certain organ, starting from the primary site. The network makes no prior assumptions as to the connections between anatomic sites, but uses the steady-state information (autopsy data) to infer their values. Figure 1 shows the distribution of lung cancer metastases from the original autopsy data set and the relative ranking of metastatic sites based on the MFPT for five primary tumors. It is clear that in lung cancer, adrenal metastases occur relatively quickly (Figure 1b) compared to other metastatic sites. In lung cancer, the short passage times and relatively high transition probabilities associated with adrenal metastases are in the range of what is typically seen for regional lymph node disease rather than for hematogenously spread sites. By comparison, adrenal metastasis in colon, breast, prostate, and other solid tumors (Figures 1c–f) do not share the short MFPT from the primary site to the adrenal.

Figure 1.

Figure 1a. Metastatic tumor distribution for primary lung cancer from DiSibio and French data set. 163 primary lung cancer autopsies; 619 metastatic tumors; 59 adrenal metastases. Histogram is normalized so that sum of areas is one.

Figure 1b. Mean first-passage times to other organs associated with a collection of random walkers leaving the primary lung tumor. Adrenal is the first visited site after lymph nodes.Times are normalized so that regional lymph nodes is unity, others are scaled to it.

Figure 1c. Mean first-passage times to other organs associated with a collection of random walkers leaving the primary breast tumor. Adrenal is the seventh visited site. Times are normalized so that regional lymph nodes is unity, others are scaled to it.

Figure 1d. Mean first-passage times to other organs associated with a collection of random walkers leaving the primary colon tumor. Adrenal is the sixth visited site.Times are normalized so that regional lymph nodes is unity, others are scaled to it.

Figure 1e. Mean first-passage times to other organs associated with a collection of random walkers leaving the primary bladder tumor. Adrenal is the sixth visited site.Times are normalized so that regional lymph nodes is unity, others are scaled to it.

Figure 1f. Mean first-passage times to other organs associated with a collection of random walkers leaving the primary prostate tumor. Adrenal is the sixth visited site. Times are normalized so that regional lymph nodes is unity, others are scaled to it.

Lymphatic theory of adrenal gland metastasis in primary lung cancer

For one to accept a lymphatic theory of adrenal gland spread several postulations need to be addressed. First, the presence of an anatomic connection between the primary tumor and the adrenal gland needs to be proven. Second, if those anatomic connections truly play a role in lung cancer spread, a preferential or earlier involvement of the adrenal gland on the ipsilateral side of the primary tumor should be seen. Lastly, patients with aggressively managed ipsilateral metastases should have improved outcomes. This is based on the presumptions that ipsilateral adrenal metastases are lymphatic and thus less advanced disease than their purely hematogenously spread contralateral counterparts. This is analogous to improved outcome in patients with loco regional nodal spread compared to distant metastatic disease.

Anatomic plausibility of direct lymphatic connections

From the lung, direct lymphatic tributaries exist to the thoracic duct and to celiac and paraaortic nodes.[6] This is confirmed by postmortem examination of coal miners showing coal dust deposition in retroperitoneal lymph nodes in 87% of cadavers suggesting a direct passage to the abdomen from lung. [7]

The thoracic duct is the main collecting lymphatic vessel of the chest, receiving lymphatic drainage from the abdominal viscera. The flow of lymph via the thoracic duct is unidirectionally upward due to the lymphatic vessel valvular structure and smooth muscle in the vessel wall. However, if one in theory interrupts the typical upward lymph flow, the cancer cells can potentially travel caudad and reach the adrenal gland via those connections. In examining the entire lengths of the thoracic duct postmortem Onuigbo found cancer cells in the thoracic duct in 62% of cases with the majority of cells clumping at the lymphatic valves, thus creating a potential for blockage.[8] No human studies have looked at the flow direction in the thoracic duct in living cancer patients.

Incidence and laterality of adrenal metastases in NSCLC

Not surprisingly, there is an increase in the incidence of adrenal involvement as a patient’s disease progresses. Clinically isolated adrenal gland metastases discovered on staging CT scans in patients otherwise presumed operable are rare in the pre PET era and are reported in 1.6% to 3.5% of NSCLC. [9, 10] The true number of adrenal metastases in this group is higher due to the poor ability of imaging technology to detect small tumor deposits. Matthews and Finke illustrated this point when reporting their autopsy series of patients dying within one month of curative lung resection. In their series 3% to 7.6% of patients were discovered to have adrenal metastases on autopsy. [11, 12] Rates of adrenal involvement increase further to 9% in living metastatic cancer patients and to 25–39% in autopsy series. [2, 13–17]

Unfortunately, laterality is rarely reported in surgical, radiation or autopsy series. Out of twenty-nine manuscripts reviewed only seven reported laterality. [2, 9–36] Only five provided data sufficient to address the relevant question, and the majority of those had small numbers of patients and were retrospective. Porte reported higher likelihood of ipsilateral involvement 72% vs. 28% in patients with solitary adrenal metastases (N=43, p<0.01).[18] Kocijancic reported on patients with both unilateral and bilateral adrenal metastases and noted a 50% ipsilaterality (N=50). The contralateral and bilateral disease was reported in conglomerate and thus it is not possible to estimate the difference between ipsi- and contralateral involvement in this study. [36] It is important to notice that the prevalence of ipsilateral involvement depends on the total burden of metastatic disease. If adrenal is the only site of metastasis, 50% of the metastases are ipsilateral. However, if adrenal metastases are present together with other sites, the typical ipsilaterality rate is numerically lower at 25% (p=0.14). In addition if patients initially operated for stage I-IIIA disease recur in the adrenal gland, 63% of recurrences are ipsilateral compared to 37% of contralateral or bilateral. This relationship is opposite in patients presenting with non-operable cancers (IIIB or IV) where incidence of ipsilateral disease is only 29% compared to 71% of contralateral/bilateral ones (p=0.034). [36] Similar patterns are reported by Porte who showed that patients with ipsilateral adrenal gland involvement have lower burden of systemic disease and are less likely to have other distant metastases than if contralateral adrenal gland is involved, at 20% vs. 62% respectively. [9]

This increased incidence of ipsilateral involvement is also confirmed in a much larger autopsy series. In patients in whom adrenal metastasis was the only metastatic site, the ipsilateral adrenal gland is 1.59 times more likely to be involved than contralateral one (61% vs. 39%, N=405, P< 0.01). In patients with more than one metastatic site, ipsilateral adrenal gland still consistently receives more metastases. This difference decreases as more metastatic sites develop and completely disappears in patients with 6 or more organs involved. In addition, bilateral adrenal metastases appear more common as disease progresses. For example, in a patient with 4 metastatic sites, the incidence of bilateral adrenal metastases is approximately 60% compared to 5% if only 2 metastatic sites are discovered. [17] Similar results were reported by Onuigbo who examined 1000 cadavers and reported 61% prevalence of ipsilateral involvement compared to 39% contralateral. [2] As a composite, this data indicates that as cancer progresses and hematogenous pathways become definitely involved as indicated by other distant sites of spread, the likelihood of circulating tumor cells establishing metastases in the contralateral adrenal gland increases. It is possible that in the beginning stages of spread cancer could be reaching the adrenal gland via the lymphatic channels before hematogenous metastases develops, thus explaining ipsilateral preponderance.

Differential outcome in aggressively managed ipsilateral vs. contralateral adrenal metastases

If managed solely with chemotherapy even isolated adrenal metastases progress to death with reported survival of only 6 to 9 months. [10, 27] No randomized clinical trials compared the survival outcome of surgically managed adrenal metastases versus chemotherapy treated ones. In a small retrospective case series by Luketich (N =14), the median survival was 31 months for patients treated with chemotherapy followed by surgical resection of both primary mass and a solitary adrenal metastasis versus 8.5 in patients treated with chemotherapy alone (P=0.03). [22] Reported outcomes are better with adrenalectomy and radiation therapy (16 – 31 months) than medically treated metastatic disease (9–12 months.)[10, 18–20, 22, 23, 29, 30, 32] If involvement of the ipsilateral adrenal gland is lymphatic in origin, one should see improvement in survival if the adrenal metastasis is managed aggressively with surgery and radiation similar to an improved survival seen in patients with stage III disease compared to patients with stage IV disease. The main challenge in looking at this question is that studies reporting outcomes of aggressively managed isolated adrenal metastases fail to report outcomes based on laterality. Two studies deserve mentioning. Porte in the pre PET era showed that after resection of solitary adrenal metastases recurrences are more common in patients with contralateral adrenal metastases than ipsilateral ones. (N= 43, 83% vs. 77% p >0.05). Raz compared survival outcomes based on the laterality of the metastasis in patients with aggressively managed adrenal metastases as defined by surgical resection together with resection of primary tumor. PET scan was not required but was performed in some patients thus giving us a purer population in regards to undetected other distant metastases. In a small cohort of patients (N=20) he showed improvement in 5-year survival from 0% to 83% in contralateral vs. ipsilateral disease. (p=0.003). [27]

Conclusions

In summary, current models as well as a review of the literature suggests there is a difference in prevalence and a possible difference in outcomes of patients with ipsilateral vs. contralateral adrenal metastasis. Current guidelines such as those from the National Comprehensive Cancer Network, however, do not recommend discrimination in the treatment of adrenal metastasis based on laterality.

The last decade has seen the development of a number of improvements for the local therapy of metastatic disease, including the development of robotic assisted surgery as well as stereotactic radiotherapy. As these technologies reduce the burden of localized treatment of metastasis, it is critical that oncologists carefully define the populations that are most likely to benefit from interventions. Randomized, prospective clinical trials of surgical and other localized treatments will be difficult to complete in this relatively rare population. The creation of mathematical models may guide optimal anatomical sites for local therapy, but these models require clinical validation of their predictive ability. We would propose that large organizations such as the cooperative groups make an effort to create a nationwide registry of local therapy applied to oligometastatic patients. Such a registry should include surgical as well as radiosurgery based approaches and then could then be mined to compare the outcome with ipsilateral and contralateral treatments of adrenal tumors as well as answer other questions that may be relevant to the oncology community. This approach may provide the kind of prospective data that would serve both to validate existing mathematical models of metastasis and shape guidelines for the surgical management of metastatic adrenal disease. If successful, this modeling and clinical validation strategy can be expanded to other tumor types.