The claim is correct for:
- Canthaxanthin is used as a colorant in foodstuff, particularly so in farmed salmon, but also eggs and poultry
- Canthaxanthin does also occur naturally, in some fish, in tiny trace amounts
- Canthaxanthin is linked to eye disease, like the aptly named "Canthaxanthin retinopathy",
- the negative effects occur at relatively higher doses, which are usually not reached when eating salmon occasionally, but frequent consumption will bring some consumers easily above the acceptable daily intake threshold
Thus the claim is going most probably into hyperbole for "most toxic food". The exact doses for the development of the retinopathy are not known.
It is concerning that the colorant is frequently overdosed in farmed fish for the resulting assumed human intake. 80 mg/kg in feedingstuff is allowed, but it is calculated that 'safety' would be achieved by limiting that to 25 mg/kg, but market demand ensures that 80mg is still applied to achieve the desired colour.
In Europe the stuff is regulated as follows:
Canthaxanthin has been authorised at Community level under EC number E 161g as a colouring matter in feedingstuffs under the conditions set out in Council Directive 70/524/EEC on additives in feedingstuffs:
Maximum content mg/kg complete feedingstuff: 80 mg/kg
Use permitted from the age of 6 months onwards
The mixture of canthaxanthin with astaxanthin is allowed provided that the total concentration of the mixture does not exceed 100 mg/kg in the complete feedingstuff.
It has also been found in bacteria (Saperstein and Starr, 1954), crustacea (Davies et al., 1970; Thommen and Wackernagel, 1964) and various species of fish including carp Cyprinus carpio (Katayama et al., 1971; 1973), golden mullet Mugil auratus, annular seabream Diplodus annularis, and trush wrasse Crenilabrus tinca (Czeczuga, 1973). Canthaxanthin is not encountered in wild Atlantic salmon but represents a minor carotenoid in wild Pacific salmon (Kitahara, 1983; 1984a,b; Matsuno et al., 1980). It was also reported in the wild trout Salmo trutta (Thommen and Gloor, 1965).
Today the concentration of carotenoids (mainly canthaxanthin and astaxanthin) exceeds 8 mg/kg of flesh and all producers try to reach a level that represents a value of 16 on the "Roche Color Card" (Torrissen, 2000). It must be noted that this scale is specific for measuring the pink colour due to astaxanthin and is not adapted to the orange hue obtained with canthaxanthin. Although a linear relationship has been established between the Color Card Score and the astaxanthin content in fish flesh, no such data is available for canthaxanthin.
The development over time of processing and storage operations, which can impact on canthaxanthin flesh concentration (see 3.4), has led to an increased quantity of pigments added to the diet to compensate for the degrading effects of processing.
In wild fish, carotenoid levels of up to 20-25 mg/kg have been reported in the flesh of trout (Storebakken and No, 1992), a value comparable to that reported for coho salmon by Schiedt et al. (1981). It must be noted that canthaxanthin, when present, is a very minor component
Canthaxanthin is the sole pigment of the carotenoid family registered for the use in both animal feeds and human foods. In the European Union, it is also the sole carotenoid pigment for which an Acceptable Daily Intake (ADI) has been established (0.03 mg canthaxanthin per kg body weight).
— European Commission, Health & Consumer Protection Directorate-General, Directorate C - Scientific Opinions, C2 - Management of scientific committees; scientific co-operation and networks
"Opinion of the Scientific Committee on Animal Nutrition on the use of canthaxanthin in feedingstuffs for salmon and trout, laying hens, and other poultry", Adopted on 17 April 2002. pdf
In that paper they show an estimated human consumption of this carotenoid is indeed from eggs, poultry and highest from farmed fish:
13.7mg/kg fish muscle, 300g of that per human and per day, yielding 4.11 mg/day, or 0.069 Canthaxanthin Daily Intake (mg/kg bw) for a 60 kg person.
The retinopathy can resolve over time, if one stops consuming it, which was previously done on purpose as a prescribed medicine, or over the counter as a tanning pill. That is now no longer allowed in many countries, but while those pills were available, the following long-term follow-up identified retinopathy patients who took 30mg a day for 2 months up to three years:
Canthaxanthin retinopathy was first described in 1982 by Cortin et al., and occurs as golden particles in the paramacular region after ingestion of high doses of canthaxanthin.
[…]
A decrease in the quantity of golden particles after termination of canthaxanthin consumption was noted upon repeat photography over a 1½-year period, and for up to 6 years, but we have demonstrated for the first time that complete disappearance of the golden particles may take more than 20 years.
— Arno Hueber & André Rosentreter & Maria Severin: "Canthaxanthin Retinopathy: Long-Term Observations", Ophthalmic Res 2011;46:103–106 DOI: 10.1159/000323813
Freely accessible:
[…] canthaxanthin retinal crystal deposition is a very common finding in patients with prolonged use of the drug. Symptomatic visual loss is less common and correlates with total dosage and possibly patient age.
— Robert A. Beaulieu et al.: "Canthaxanthin Retinopathy with Visual Loss: A Case Report and Review", Case Reports in Ophthalmological Medicine, 2013. doi
For the generalised title of the question, "is it toxic to human health", again the dosage makes a poison:
At high doses, canthaxanthin has caused a serious, potentially fatal blood disorder called aplastic anemia. Canthaxanthin can also cause diarrhea, nausea, stomachcramps, dry and itchy skin, hives, orange or red body secretions, and other side effects.
— WebMD: Canthaxanthin - Uses, Side Effects, and More
Relying mainly on this fatal outcome:
The drug's present means of distribution makes monitoring for toxic effects difficult. Thus, the frequency of adverse effects associated with canthax- anthin use, such as bone marrow suppression, is unknown. Even if there is only a small risk of these toxic effects, the use of the drug for cosmetic purposes does not justify this risk.
A 20-year-old white woman had a 4-month history of malaise, headaches, […]
She reported that about 4 months prior to being hospitalized she took a course of "tanning pills" containing canthaxanthin, which turned her deep orange. […] the dose was not known […]
15 days after admission, at which time the hematocrit was 0.08 and the platelet count was 11.0x109L, the patient was discharged to her home at her request for what was expected to be terminal care. She died 48 hours later.
— Renata Bluhm et al.: "Aplastic Anemia Associated With Canthaxanthin Ingested for 'Tanning' Purposes", JAMA, 1990; 264 (9): p1141–1142.
