The intrusive complexes of Gremiakha-Vyrmes and Soustov represent the two extremes of the Early Proterozoic alkaline plutons of Kola, predominantly composed of feldespathoidal syenites. Gremaikha-Vyrmes rocks (zircon age: 1,884±6 Ma) have trace-element and isotope signatures (⁸⁷Sr/⁸⁶Sr₁-e7 0.704, _eNd₁ ≈ -3-1.3) compatible with an ultimate mantle origin. Soustov syenites (zircon age: 1,872 ≠ 8 Ma) are totally different and show an acute crustal imprint. They have sodaline and analcite instead of nepheline, contain a plethora of REE-HFSE-rich accessories, and are characterised by elevated contents of F, CI, REE, Y, Th, U, Zr, Hf, Nb, Ta, Sn, Be, Li, Rb, Ti, Pb and Cs, negative Eu anomalies, K/Rb ≈ 190-160, Nd/Th ≈ 3, and Nb/Ta ≈ 12, with extremely high ⁸⁷Sr/⁸⁶Sr₁ (>0.720) and, at the same time, relatively high εNd_t (≈ -1.6-1.7). In this paper, we explore the idea that the anomalous features of Soustov syenites can be explained if we assume they are derived from a matasomatic agent, initially an H₂O-CO₂ supercritical fluid released by alkaline mafic magmas, that was profoundly contaminated during percolation through crustal materials. As percolation advanced, the bulk composition of the fluid solute changed from alkali halides and carbonates to a silica-undersaturated alkaline melt. When the fluid cooled to a temperature of 550-600 °C, it reached the point at which vapor and melt were no longer miscible and split into two components, a vapour phase and a CI- and F-rich silics-undsersaturated silicate melt that crystallised to produce Soustov syenites. To study this process, we have developed a numerical method for modelling the solute composition of the fluid during the infiltration metasomatism. Our results, using the LREE abundances and the Sr and Nd isotope composition of a Gremiakha-Vyrmes pegmatite as the starting solute composition of the fluid, and the mode and mineral trace-element and isotope composition of a common Kola gneiss as representative of percolated materials, indicate that the fluid would have acquired a signature closely matching Soustov's even in the case of Nd isotopes, if the gneiss age is 2.9 Ga, near its real age. This model is still a mere working hypothesis that needs further refinements, but may represent a reasonable explanation of the genesis of anomalous alkaline rocks with high ⁸⁷Sr/⁸⁶Sr_t and εNd_t ≥ 0, either saturated or undersaturated, which are difficult to understand in terms of magmatic fractionation/contamination.