Yes it will impact the forecasting. there could be two way out. firstly you can include the dummy variable at the break point in VAR model. if you are running ARIMA then you will have consider your data without break.
For possible structural break Gregory Hansan and Hatemi-J models estimation may be also be explored followed by forecasting. Presence of structural break may be one of the aspects that may be present in the relationship among the variables you are interested in. In that case way may like to understand non-linear cointegration techniques (NARDL, QNARDL) - for better forecasting.