Machine learning is a powerful tool to study the effect of cancer on species and ecosystems

Authors: Antoine M. Dujon, Marion Vittecoq, Georgina Bramwell, Frédéric Thomas, Beata Ujvari

Source: METHODS IN ECOLOGY AND EVOLUTION (AUG 2021)

Abstract

Cancer is an understudied but important process in wildlife. Cancerous cells are proposed to have had significant effect on the evolution of metazoan species due to their negative effect on host fitness. However, gaining knowledge on the impact of cancer on species and ecosystems is currently relatively slow as it requires expertise in both ecology and oncology. The field can greatly benefit from automation to reduce the need of excessive manpower and analyse complex ecological datasets.

In this commentary, we examine how machine learning has been used to gain knowledge on oncogenic processes in wildlife. Using a landscape ecology approach, we explore spatial scales ranging from the size of a molecule up to whole ecosystems and detail, for each level, how machine learning has been used, or could contribute to obtain insights on cancer in wildlife populations and ecosystems.

We illustrate how machine learning is a powerful toolbox to conduct studies at the interface of ecology and oncology. We provide guidance for the readers of both fields on how to implement machine learning tools in their research and identify directions to move the field forward using this promising technology. We demonstrate how applying machine learning to complex ecological datasets will (a) contribute to quantitating the effect of cancer at different life stages in wildlife; (b) allow the mining of long-term datasets to understand the spatiotemporal variability of cancer risk factors and (c) contribute to mitigating cancer risk factors and the conservation of endangered species.

With this study, we aim to facilitate the use of machine learning to wildlife species and to encourage discussion between the scientists of the fields of oncology and ecology. We highlight the importance of international and pluridisciplinary collaborations to collect high-quality datasets on which efficient machine learning algorithms can be trained.

Tumors (re)shape biotic interactions within ecosystems: Experimental evidence from the freshwater cnidarian Hydra

Authors: Justine Boutry; Juliette Mistral; Laurent Berlioz; Alexander Klimovich; Jácint Tökölyi; Laura Fontenille; Beata Ujvari; Antoine M. Dujon; Mathieu Giraudeau; Frédéric Thomas

Source: Science of The Total Environment (AUG 2021)

Abstract

While it is often assumed that oncogenic processes in metazoans can influence species interactions, empirical evidence is lacking.

Here, we use the cnidarian Hydra oligactis to experimentally explore the consequences of tumor associated phenotypic alterations for its predation ability, relationship with commensal ciliates and vulnerability to predators.

Unexpectedly, hydra’s predation ability was higher in tumorous polyps compared to non-tumorous ones. Commensal ciliates colonized preferentially tumorous hydras than non-tumorous ones, and had a higher replication rate on the former.

Finally, in a choice experiment, tumorous hydras were preferentially eaten by a fish predator. This study, for the first time, provides evidence that neoplastic growth has the potential, through effect(s) on host phenotype, to alter biotic interactions within ecosystems and should thus be taken into account by ecologists.

Graphical abstract

Is There One Key Step in the Metastatic Cascade?

Authors: Antoine M. Dujon, Jean-Pascal Capp, Joel S. Brown, Pascal Pujol, Robert A. Gatenby, Beata Ujvari, Catherine Alix-Panabières and Frédéric Thomas

Source: Evolutionary Applications (MAY 2021)

Brief summary of the paper:

Simple Summary

To successfully metastasize, cancer cells must complete a sequence of obligatory steps called the metastatic cascade. To model the metastatic cascade, we used the framework of the Drake equation, initially created to describe the emergence of intelligent life in the Milky way, using a similar logic of a sequence of obligatory steps.

Then within this framework, we used simulations on breast cancer to investigate the contribution of each step to the metastatic cascade.

We show that the half-life of circulating tumor cells is one of the most important parameters in the cascade, suggesting that therapies reducing the survival of those cells in the vascular system could significantly reduce the risk of metastasis.

Abstract

The majority of cancer-related deaths are the result of metastases (i.e., dissemination and establishment of tumor cells at distant sites from the origin), which develop through a multi-step process classically termed the metastatic cascade.

The respective contributions of each step to the metastatic process are well described but are also currently not completely understood. Is there, for example, a critical phase that disproportionately affects the probability of the development of metastases in individual patients?

Here, we address this question using a modified Drake equation, initially formulated by the astrophysicist Frank Drake to estimate the probability of the emergence of intelligent civilizations in the Milky Way. Using simulations based on realistic parameter values obtained from the literature for breast cancer, we examine, under the linear progression hypothesis, the contribution of each component of the metastatic cascade. Simulations demonstrate that the most critical parameter governing the formation of clinical metastases is the survival duration of circulating tumor cells (CTCs).

COVID-19 disruption reveals mass-tourism pressure on nearshore sea turtle distributions and access to optimal breeding habitat

Authors: Gail Schofield, Liam C. D. Dickson, Lucy Westover, Antoine M. Dujon & Kostas A. Katselidis

Source: Evolutionary Applications (MAY 2021)

Brief summary of the paper:

Quantifying the extent to which animals detect and respond to human presence allows us to identify pressure (disturbance) and inform conservation management objectively; however, obtaining baselines against which to compare human impact is hindered in areas where human activities are already well established.

For example, Zakynthos Island (Greece, Mediterranean) receives around 850,000 visitors each summer, while supporting an important loggerhead sea turtle rookery (~300 individuals/season).

The coronavirus (COVID-19)-driven absence of tourism in May–June 2020 provided an opportunity to evaluate the distribution dynamics of this population in the absence (2020) vs. presence (2018 and 2019) of visitors using programmed unmanned aerial system (UAS) surveys.

Ambient sea temperature transitioned from suboptimal for breeding in May to optimal in late June, with turtle distribution appearing to shift from shallow (to benefit from waters 3–5°C above ambient) to deeper waters in 2018 and 2019, but not 2020. The 2020 data set demonstrated that increased tourism pressure, not temperature, drives turtles offshore.

Specifically, >50% of turtles remained within 100 m of shore at densities of 25–50 visitors/km, even when sea temperature rose, with 2018 and 2019 data supporting this trend. Reduced access to warmer, nearshore waters by tourism could delay the onset of nesting and increase the length of the egg maturation period between nesting events (internesting interval) at this site. A coastal refuge zone could be delimited in May–June where touristic infrastructure is minimal, but also where turtles frequently aggregate.

In conclusion, sea turtles appear capable of perceiving changes in the level of human pressure at fine spatial and temporal scales and adjusting their distribution accordingly.

Genetic structure and gene flow in the Flame Robin (Petroica phoenicea)

Authors: Beckmann, C; Major, RE; Frankham, GJ; Thomas, S; Biro, PA; Ujvari, B; Neaves, L

Source: EMU-AUSTRAL ORNITHOLOGY (MAY 2021)

Brief summary of the paper:

Robins in the family Petroicidae are characteristic of the woodland bird community that is threatened in Australia as a result of habitat loss and fragmentation. Flame Robin (Petroica phoenicea) populations declined by 56% between 1980 and 2000, with habitat loss likely being the primary cause.

Given that Flame Robins primarily breed at high elevation, populations may become more isolated due to anthropogenic change, resulting in increased inbreeding and loss of genetic diversity that may accelerate local extinction.

We estimated the genetic structure and recent gene flow among four populations (n = 70 birds) of this vulnerable (NSWSC) species across a 670 km portion of its range in temperate south-eastern Australia using 14 genetic markers. We found no significant differences in genetic diversity amongst populations and little population structuring – only the northernmost population showing a weak signal of differentiation. However, we detected little recent migration between the northern and southern sites, possibly due to recent fragmentation.

We conclude that habitat loss is a conservation concern for this Vulnerable species and further work and ongoing genetic monitoring is needed, particularly given high elevation breeding sites that are vulnerable in the face of a changing climate.

Does Cancer Biology Rely on Parrondo’s Principles?

Authors: Jean-Pascal Capp, Aurora M Nedelcu, Antoine M Dujon, Benjamin Roche, Francesco Catania, Beata Ujvari, Catherine Alix-Panabières and Frédéric Thomas

Source: CANCERS (MAY 2021)

Brief summary of the paper:

Many aspects of cancer biology remain puzzling, including the proliferative and survival success of malignant cells in spite of their high genetic and epigenetic instability as well as their ability to express migrating phenotypes and/or enter dormancy despite possible fitness loss.

Understanding the potential adaptive value of these phenotypic traits is confounded by the fact that, when considered separately, they seem to be rather detrimental at the cell level, at least in the short term. Here, we argue that cancer’s biology and success could frequently be governed by processes underlying Parrondo’s paradox, whereby combinations of intrinsically losing strategies may result in winning outcomes.

Oncogenic selection would favor Parrondo’s dynamics because, given the environmental adversity in which malignant cells emerge and evolve, alternating between various less optimal strategies would represent the sole viable option to counteract the changing and deleterious environments cells are exposed to during tumorigenesis.

We suggest that malignant processes could be viewed through this lens, and we discuss how Parrondo’s principles are also important when designing therapies against cancer.

A review of the potential effects of climate change on disseminated neoplasia with an emphasis on efficient detection in marine bivalve populations

Authors: Georgina Bramwell, Aaron G. Schultz, Craig D. H. Sherman, Mathieu Giraudeau, Frédéric Thomasb, Beata Ujvari, Antoine M Dujon

Source: SCIENCE OF THE TOTAL ENVIRONMENT (FEB 2021)

Brief summary of the paper:

Climate change not only directly impacts marine environments by shifting water temperatures, salinity, pH and dissolved oxygen concentrations, but may also indirectly contribute to the emergence of additional ecosystem stressors, such as infectious diseases, including bivalve disseminated neoplasia.

Disseminated neoplasia, a form of cancer found in some bivalves – recently discovered to be transmissible in at least six species – has been shown to impair bivalve health and fitness, with occasional mass outbreaks causing high levels of mortality. As the ability of the host bivalve to respond to disseminated neoplasia, and the survival and transmissibility of disseminated neoplasia both depend on environmental factors, it is crucial to understand the interaction between climate change and disseminated neoplasia epidemiology.

Furthermore, with bivalves being species of high ecological and economic importance, there is a rising need for the development of efficient disseminated neoplasia diagnostic tools in order to detect, mitigate and potentially prevent deleterious disseminated neoplasia outbreaks.

Therefore, in this study, we reviewed the current knowledge of climate impacted environmental parameters on disseminated neoplasia and identified best practices and methodology for the detection of transmissible disseminated neoplasia in the wild.

By exploring the potential effects changing climate has on disseminated neoplasia dynamics, we identified future research directions in order to advance the field. This included using state of the art disease detection methods and taking into account species’ ecological niches to understand the dynamic of disseminated neoplasia outbreaks in the wild and to investigate whether disseminated neoplasia is present in freshwater ecosystems.

Finally, we provided a comprehensive step-by-step guideline for an evidence-based detection of this disease in marine ecosystems.

Group phenotypic composition in cancer

Authors: Jean-Pascal Capp, James DeGregori, Aurora M Nedelcu, Antoine M Dujon, Justine Boutry, Pascal Pujol, Catherine Alix-Panabières, Rodrigo Hamede, Benjamin Roche, Beata Ujvari, Andriy Marusyk, Robert Gatenby, Frédéric Thomas

Source: eLife (MAR 2021)

Brief summary of the paper:

Although individual cancer cells are generally considered the Darwinian units of selection in malignant populations, they frequently act as members of groups where fitness of the group cannot be reduced to the average fitness of individual group members.

A growing body of studies reveals limitations of reductionist approaches to explaining biological and clinical observations. For example, induction of angiogenesis, inhibition of the immune system, and niche engineering through environmental acidification and/or remodeling of extracellular matrix cannot be achieved by single tumor cells and require collective actions of groups of cells. Success or failure of such group activities depends on the phenotypic makeup of the individual group members.

Conversely, these group activities affect the fitness of individual members of the group, ultimately affecting the composition of the group. This phenomenon, where phenotypic makeup of individual group members impacts the fitness of both members and groups, has been captured in the term ‘group phenotypic composition’ (GPC).

We provide examples where considerations of GPC could help in understanding the evolution and clinical progression of cancers and argue that use of the GPC framework can facilitate new insights into cancer biology and assist with the development of new therapeutic strategies.

Machine learning to detect marine animals in UAV imagery: effect of morphology, spacing, behaviour and habitat

Authors: Antoine M. Dujon; Daniel Ierodiaconou; Johanna J. Geeson; John P. Y. Arnould; Blake M. Allan; Kostas A. Katselidis; Gail Schofield

Source: Remote Sensing in Ecology & Conservation (MAY 2021)

Brief summary of the paper:

Machine learning algorithms are being increasingly used to process large volumes of wildlife imagery data from unmanned aerial vehicles (UAVs); however, suitable algorithms to monitor multiple species are required to enhance efficiency.

Here, we developed a machine learning algorithm using a low‐cost computer. We trained a convolutional neural network and tested its performance in: (1) distinguishing focal organisms of three marine taxa (Australian fur seals, loggerhead sea turtles and Australasian gannets; body size ranges: 0.8–2.5 m, 0.6–1.0 m, and 0.8–0.9 m, respectively); and (2) simultaneously delineating the fine‐scale movement trajectories of multiple sea turtles at a fish cleaning station.

For all species, the algorithm performed best at detecting individuals of similar body length, displaying consistent behaviour or occupying uniform habitat (proportion of individuals detected, or recall of 0.94, 0.79 and 0.75 for gannets, seals and turtles, respectively). For gannets, performance was impacted by spacing (huddling pairs with offspring) and behaviour (resting vs. flying shapes, overall precision: 0.74).

For seals, accuracy was impacted by morphology (sexual dimorphism and pups), spacing (huddling and creches) and habitat complexity (seal sized boulders) (overall precision: 0.27). For sea turtles, performance was impacted by habitat complexity, position in water column, spacing, behaviour (interacting individuals) and turbidity (overall precision: 0.24); body size variation had no impact.

For sea turtle trajectories, locations were estimated with a relative positioning error of <50 cm. In conclusion, we demonstrate that, while the same machine learning algorithm can be used to survey multiple species, no single algorithm captures all components optimally within a given site.

We recommend that, rather than attempting to fully automate detection of UAV imagery data, semi‐automation is implemented (i.e. part automated and part manual, as commonly practised for photo‐identification). Approaches to enhance the efficiency of manual detection are required in parallel to the development of effective implementation of machine learning algorithms.

Linking pollution and cancer in aquatic environments: A review

Authors: Ciara Baines; Adelaide Lerebours; Frederic Thomas; Jerome Fort; Randel Kreitsberg; Sophie Gentes; Richard Meitern; Lauri Saks; Beata Ujvari; Mathieu Giraudeau; Tuul Seppa

Source: ENVIRONMENT INTERNATIONAL (JAN 2021)

Brief summary of the paper:

Due to the interconnectedness of aquatic ecosystems through the highly effective marine and atmospheric transport routes, all aquatic ecosystems are potentially vulnerable to pollution. Whilst links between pollution and increased mortality of wild animals have now been firmly established, the next steps should be to focus on specific physiological pathways and pathologies that link pollution to wildlife health deterioration.

One of the pollution-induced pathologies that should be at the centre of attention in ecological and evolutionary research is cancer, as anthropogenic contamination has resulted in a rapid increase of oncogenic substances in natural habitats. Whilst wildlife cancer research is an emerging research topic, systematic reviews of the many case studies published over the recent decades are scarce.

This research direction would (1) provide a better understanding of the physiological mechanisms connecting anthropogenic pollution to oncogenic processes in non-model organisms (reducing the current bias towards human and lab-animal studies in cancer research), and (2) allow us to better predict the vulnerability of different wild populations to oncogenic contamination. This article combines the information available within the scientific literature about cancer occurrences in aquatic and semi-aquatic species.

For the first aim, we use available knowledge from aquatic species to suggest physiological mechanisms that link pollution and cancer, including main metabolic detoxification pathways, oxidative damage effects, infections, and changes to the microbiome.

For the second aim, we determine which types of aquatic animals are more vulnerable to pollution-induced cancer, which types of pollution are mainly associated with cancer in aquatic ecosystems, and which types of cancer pollution causes.

We also discuss the role of migration in exposing aquatic and semi-aquatic animals to different oncogenic pollutants. Finally, we suggest novel research avenues, including experimental approaches, analysis of the effects of pollutant cocktails and long-term chronic exposure to lower levels of pollutants, and the use of already published databases of gene expression levels in animals from differently polluted habitats.